(19)
(11)EP 1 952 207 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
29.07.2020 Bulletin 2020/31

(21)Application number: 06839841.1

(22)Date of filing:  10.11.2006
(51)International Patent Classification (IPC): 
G05B 13/02(2006.01)
H02M 1/00(2006.01)
H02M 3/157(2006.01)
H02M 3/335(2006.01)
(86)International application number:
PCT/US2006/060803
(87)International publication number:
WO 2007/059447 (24.05.2007 Gazette  2007/21)

(54)

METHODS AND SYSTEMS FOR ADAPTIVE CONTROL

VERFAHREN UND SYSTEME ZUR ADAPTIVEN STEUERUNG

PROCEDES ET SYSTEMES DE COMMANDE ADAPTATIVE


(84)Designated Contracting States:
AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LI LT LU LV MC NL PL PT RO SE SI SK TR

(30)Priority: 11.11.2005 US 735384 P
27.10.2006 US 553917

(43)Date of publication of application:
06.08.2008 Bulletin 2008/32

(73)Proprietor: L&L Engineering LLC
Lee, NH 03824 (US)

(72)Inventors:
  • LATHAM, Paul
    Lee, NH 03824 (US)
  • CANFIELD, John, C.
    Lee, NH 03824 (US)

(74)Representative: FRKelly 
27 Clyde Road
Dublin D04 F838
Dublin D04 F838 (IE)


(56)References cited: : 
US-A- 5 475 628
US-A1- 2005 231 871
US-A1- 2003 174 005
  
  • TSANG K M ET AL: "Adaptive control of power factor correction converter using nonlinear system identification", IEE PROCEEDINGS: ELECTRIC POWER APPLICATI, INSTITUTION OF ELECTRICAL ENGINEERS, GB, vol. 152, no. 3, 8 April 2005 (2005-04-08) , pages 627-633, XP006023957, ISSN: 1350-2352, DOI: 10.1049/IP-EPA:20045058
  • SMITH S ED - SMITH S W: "Digital Signal Processing: A Practical Guide for Engineers and Scientists", [DEMYSTIFYING TECHNOLOGY SERIES], NEWNES, 1 January 2003 (2003-01-01), XP040425435, ISBN: 978-0-7506-7444-7
  • MAKSIMOVIC D ET AL: "System Identification of Power Converters With Digital Control Through Cross-Correlation Methods", IEEE TRANSACTIONS ON POWER ELECTRONICS, INSTITUTE OF ELECTRICAL AND ELECTRONICS ENGINEERS, USA, vol. 20, no. 5, 1 September 2005 (2005-09-01), pages 1093-1099, XP011138682, ISSN: 0885-8993, DOI: 10.1109/TPEL.2005.854035
  • DERKSEN S ET AL: "Load current estimation for control algorithms in buck converter", POWER ELECTRONICS AND APPLICATIONS, 2005 EUROPEAN CONFERENCE ON DRESDEN, GERMANY 11-14 SEPT. 2005, PISCATAWAY, NJ, USA,IEEE, PISCATAWAY, NJ, USA, 11 September 2005 (2005-09-11), pages P.1-P.10, XP010933332, DOI: 10.1109/EPE.2005.219754 ISBN: 978-90-75815-09-2
  
Note: Within nine months from the publication of the mention of the grant of the European patent, any person may give notice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed in a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).


Description

BACKGROUND



[0001] Applying digital methods to the control of systems bears the promise of creating new features, improving performance, providing greater product flexibility, and providing a lower cost. System operating characteristics dictated by a stored program, rather than the parameters of a set of discrete components, can result in cost and space savings as well as capacity for real time adaptation of those characteristics, greater sophistication in control algorithms and the ability to generate, store and recall valuable real-time functional data.

[0002] However, digital feedback control requires high resolution and high speed. These requirements have limited the adoption of digital control in many fields. The advent of low cost logic has it made possible the application of digital control techniques to cost sensitive fields. As the cost of digital logic decreases, new opportunities arise.

[0003] A typical digitally controlled feedback system has an analog to digital converter, digital loop compensator, power device driver, and an external system to be controlled. An example of a system in which application of digital control can improve performance or lower cost is the switching power supply or DC-to-DC converter. (However, many other systems would also benefit from application of digital control.)

[0004] It is very desirable to minimize the cost, size and power dissipation of a low-cost off-line switching power supply for low power applications, such as recharging cells and batteries used in portable consumer appliances, such as entertainment units, personal digital assistants, and cell phones, for example.

[0005] A PWM switched power supply requires a variable pulse width that is controlled by an error signal derived by comparing actual output voltage to a precise reference voltage. The pulse width of the switching interval must also be constrained to be within a minimum and maximum duration. These constraints arc imposed for correct PWM power supply or motor driver operation.

[0006] An example of a digitally controlled system is shown in Fig. 1.. In the example shown in Fig. 1, the system is a simple buck Switching power supply. The fundamental components are the same for any Switching power supply. The sample system shown in Fig. 1 includes three major components: a compensator preceded by an ADC, PWM and power switches, and passive LC network.

[0007] Most power management design is based on simple, continuous compensation using frequency domain analysis. Bode analysis is frequently used as the design technique of choice.

[0008] More modern techniques such as modeling of the converter in discrete time and using pole placement or optimization techniques to set the gains ate not usually considered. Recently developed digital power management chips use the digital equivalent of analog continuous lime designs. The design procedure starts with an analog prototype which is discretized and implemented in hardware.

[0009] TSANG KM ET AL, IEE PROCEEDINGS: ELECTRIC POWER APPLICATIONS, INSTITUTION OF ELECTRICAL ENGINEERS, GB, Vol. 152, No. 3, 8 April 2005, pages 627-633, XP006023957, discusses: "Adaptive control of power factor correction converter using nonlinear system identification."

[0010] SMITH SW in DEMYSTIFYING TECHNOLOGY SERIES, NEWNES, 1 January 2003, XP040425435, discusses: "Digital Signal Processing: A Practical Guide for Engineers and Scientists."

[0011] US2003174005 describes a CMOS digital PWM controller chip including a clock generator for directly generating a sampling clock at a frequency higher than a control pulse rate without requiring a phase locked loop, an oversampling analog-to-digital converter clocked by the sampling clock for converting error signals into filtered digital values, an output for controlling duty cycle of an electrical device in accordance with width-modulated digital control pulses supplied at the control pulse rate, and, digital control logic for generating the width-modulated digital control pulses in relation to the digital values. The digital control logic may include a linear feedback shift register to pseudo-randomize the digital control pulses to reduce electromagnetic interference.

[0012] MAKSIMOVIC D ET AL in IEEE TRANSACTIONS ON POWER ELECTRONICS, INSTITUTE OF ELECTRICAL AND ELECTRONICS ENGINEERS, USA, Vol. 20, No. 5, 1 September 2005, pages 1093-1099, XP011138682, discusses: "System Identification of Power Converters With Digital Control Through Cross-Correlation Methods."

[0013] DERKSEN S ET AL in POWER ELECTRONICS AND APPLICATIONS, 2005 EUROPEAN CONFERENCE ON DRESDEN, GERMANY 11-14 SEPT. 2005, PISCATAWAY, NJ, USA, IEEE, PISCATAWAY, NJ, USA, 11 September 2005, pages P.1-P.1 0, XP01 0933332 discusses: "Load current estimation for control algorithms in buck converter."

BRIEF SUMMARY



[0014] A first aspect of the present invention provides a digital controller as recited in claim 1.

[0015] A second aspect of the present invention provides a method for digital control of a system as recited in claim 7. The present disclosure provides a controller as detailed in claim 1 and a method in accordance with claim 7. Advantageous features are provided in dependent claims

[0016] In the following description, the term "embodiment" may have been used for subject-matter that is not part of the invention as defined by the appended claims.

[0017] Only those examples that comprise all the features of the independent claims are part of the invention and thus embodiments of the invention.

[0018] Parts of the subject-matter of the description not covered by the claims constitute background art or examples useful for understanding the invention.

[0019] For a better understanding of these teachings, together with other and further needs thereof, reference is made to the accompanying drawings and detailed description and its scope will be pointed out in the appended claims.

BRIEF DESCRIPTION OF THE DRAWINGS



[0020] 

Figure 1 is a graphical schematic presentation of a conventional system;

Figure 2 is a graphical schematic presentation of an embodiment of the system of these teachings;

Figure 3 is a graphical schematic representation of another embodiment of the system of these teachings;

Figure 4 shows a graphical schematic representation of a further embodiment of the system of these teachings;

Figure 5 shows a graphical schematic representation of a component of an embodiment of the system of these teachings;

Figure 6 depicts a graphical representation of another embodiment of the component of an embodiment of the system of these teachings;

Figure 7 is a graphical schematic representation of yet another embodiment of a system of these teachings;

Figure 8 depicts a graphical schematic representation of still a further embodiment of the system of these teachings;

Figures 9-14 depict graphical representations of results from embodiments of the system of these teachings;

Figure 15 is a graphical schematic presentation of a state space estimator where, in the most general formulation, the signals are vectors and the gains are matrices;

Figure 16 is a graphical representations of results showing the response of the state estimator for a Switching power supply where the state variables are capacitor voltage and inductor current;

Figure 17a is a graphical schematic representation of another estimator structure embodiment of these teachings;

Figure 17b is a graphical schematic representation of the another estimator structure embodiment of these teachings applied to of a buck regulator;

Figure 18 is a graphical representations of results obtained using the embodiment of Figure 17b;

Figure 19 is a graphical schematic representation of yet another estimator structure embodiment of these teachings; and

Figure 20 is a graphical representations of results obtained using the embodiment of Figure 19.


DETAILED DESCRIPTION



[0021] In one embodiment, the system of these teachings includes a Switching power supply, an adaptive plant estimation component capable of receiving an output voltage from the Switching power supply and an input to the Switching power supply and of providing a model of the Switching power supply; the model reflecting changes in output voltage state of the Switching power supply, a compensator design component capable of receiving the model of the Switching power supply and of providing compensator parameters, the compensator parameters reflecting changes in output voltage state of the Switching power supply, an adaptive compensator capable of receiving the compensator parameters and of providing the input control signal to the driver component. The Switching power supply includes a circuit having at least two reactive components configured to provide an output voltage and capable of being switched from one output voltage state to another output voltage state, a switching component capable of switching said circuit between one output voltage state and the other output voltage state, and a driver component capable of receiving an input control signal and of diving the switching component in order to cause switching between the one output voltage state and the other output voltage state in response to the input control signal.

[0022] Figure 2 presents a block diagram of an embodiment of the system of these teachings. Referring to Figure 2, a Switching power supply 110 includes an inductor 112 and a capacitor 114 and a switching component 116 and a driver component 118 (a PWM). The voltage across the capacitor 114 is the output voltage of the Switching power supply 110. An adaptive plant estimation component 120 receives the output voltage 124 of the Switching power supply 110 and the input control signal 126 to the Switching power supply 110 and provides a model of the Switching power supply 110, where the model reflects changes in the Switching power supply 110. The model to Switching power supply 110 is provided to a compensator design component 130, which provides compensator component parameters to the adaptive compensator 140. The compensator design parameters reflect changes in the compensation needed to account for changes in the Switching power supply 110. In some embodiments, a disturbance injection component 150 provides a substantially small noise signal that can aid in the detection of changes in the Switching power supply 110. In some embodiments, the adaptive plant estimator component 120 utilizes the LMS algorithm in order to provide a model of the Switching power supply 110. (For a description of the LMS algorithm, sec, for example, S. Haykin, Introduction to Adaptive Filters, ISBN 0-02-949460-5, pp.108-110.). In other embodiments, the adaptive plant estimator component 120 utilizes and RLS algorithm (for a description of the RLS algorithm, see, for example, S. Haykin, Introduction to Adaptive Filters, ISBN 0-02-949460-5, pp. 139-143).

[0023] It should be noted that while the exemplary Switching power supply embodiment of these teachings is described by one exemplary type, other power supply architectures such as boost, buck-boost, flyback, forward, etc are within the scope of these teachings.

[0024] In one embodiment, the controller of these teachings includes a sampling component capable of sampling an output signal from a system and an input signals from the system at a first sampling rate, the first sampling rate being at least equal to a predetermined operating rate, an input parameter obtaining component capable of receiving the output signal and the input signal sampled at the first sampling rate and of obtaining values for a plurality of input parameters, the values for the input parameters being sampled at the first sampling rate, a decimator component capable of receiving the values for the input parameters sampled at the first sampling rate and of providing subsampled values for the input parameters, subsampled values being sampled at a second sampling rate, the second sampling rate been slower than the first sampling rate, an adaptive plant estimator component capable of receiving the subsampled values of the input parameters and of obtaining a model of the system, the model reflecting variations in the system.

[0025] Although the embodiments described hereinbelow are described in terms of a particular controlled: component, it should be noted that the embodiments described hereinbelow can be applied to a wide range of other controlled components.

[0026] Figure 3 shows a block diagram representation of an embodiment of the controller of these teachings. Referring to Figure 3, the embodiment shown therein includes a sampling component 220 that samples an output signal from a system 210 and an input signal from the system at a first sampling rate, the first sampling rate being greater than or equal to a predetermined operating rate, an input parameter obtaining component 230 capable of receiving the output signal and the input signal sampled at the first sampling rate and of obtaining values for a number of input parameters, the values for the input parameters being sampled at the first sampling rate, a decimator component 240 capable of receiving the values for the input parameters sampled at the first sampling rate and of providing subsampled values for the input parameters, the subsampled values being sampled at a second sampling rate, the second sampling rate been slower than the first sampling rate, an adaptive plant estimator component 250 that receives the subsampled values of the input parameters and obtains a model of the system 210, the model reflecting variations in the system and a compensator design component 260 that receives the model of the system and of providing compensator parameters, the compensator parameters reflecting changes in the system; values of said compensator parameters being sampled at the second sampling rate; said compensator design component being capable of providing said values of said compensator parameter to a compensator 270. The compensator 270 operates at the predetermined operating rate.

[0027] In one exemplary embodiment, these teachings not being limited to that exemplary embodiment, parameters of the system 210 (DC-to-DC power supply) vary slowly. Therefore it is possible to make the parameter updates a slower, offline computation. In a linear compensator type design, an analog-to-digital converter 220 (ADC) measures the output and input (and intermediate in some embodiments) voltages in the power supply 210 and provides them to the compensator 270. This allows for both error feedback and correction of input supply variations. The ADC results are also used by the auto- and cross-correlators 230 to measure the behavior of the power supply 210. The resulting correlation coefficients are used to design the compensator. The parameter computation and compensator design are done offline at a lower sampling rate. This lowers the cost of those tasks, because the digital logic can be in the form of a simple micro-sequencer. If it is desired, the compensator can also be implemented in analog form, with digital adjustments made during the compensator design stage.

[0028] In another embodiment, shown in Figure 4, the controller of these teachings also includes a load current estimating component 280 capable of receiving the output signal sampled at the first sampling rate and state estimation data from the adaptive plant estimator component 250 and of providing estimated load current data at the first sampling rate and another decimator component 288 capable of receiving the estimated load current data at the first sampling rate and of providing estimated load current data at the predetermined operating rate to the compensator 270.

[0029] In one instance, the ADC provides inputs to a state estimator 284. The estimated states are then used by the feedback gain matrix in the compensator 270 to complete the feedback system. In another instance, a load current estimator is also included. The load current estimator 286 allows for the effect of variations in the load current to be minimized. The values from the load current estimator 286 and the state estimator 284 are provided to another decimator 288 that provides estimated load current data at the predetermined operating rate to the compensator 270.

[0030] In one instance, as shown in Figure 5, the ADC is an oversampling ADC, aDelta Sigma ADC 290 in the embodiment shown, including a an oversampling modulator, delta-sigma modulator 294 in the embodiment shown, and a decimation filter 296. The oversampling modulator 290 converts the analog input voltage into a high-speed, digital bit stream. The digital bit stream may be as little as one bit wide. Because the digital bit stream is sampled at a very high rate, it is possible to low-pass filter the bit stream and to recover a high-precision, lowcr-sample-rate representation of the analog signal.

[0031] In one embodiment, shown in Figure 6, the sampling component (220, Figure 3) is a oversampling (sigma delta in one embodiment) modulator 310 and the first sampling rate is an oversampled rate. In the embodiment shown in figure 6, the input parameter obtaining component (230, Figure 3) is an autocorrelation and crosscorrelation estimator 320. It should be noted that other embodiments of the input parameter obtaining component are possible and within the scope of these teachings. It should also be noted that embodiments are possible in which the oversampling (sigma delta in one embodiment) modulator 310 provides inputs to the state estimator 284 and the load estimator 286.

[0032] In many applications, including the DC-to-DC converter application, and in particular for embodiment utilizing the cross- and autocorrelation functions, the decimation filter (decimator) function 240 can be built-in. This reduces the cost because a one-bit multiplier is just a single gate, while a high-precision digital multiplier can be a costly design.

[0033] Figure 7 shows an embodiment of the system of these teachings in which the oversampling (sigma delta in one embodiment) modulator 310 is used as a sampling component. In the embodiment shown in Figure 7, the system is a switching power supply or a generic power supply. The driver component in the switching power supply is typically a PWM (pulse width modulator). Of specific interest in many instances, especially, but not limited to, the instance in which the driver component is a PWM, it is the specific embodiment in which the first sampling rate (or the oversampled rate in one instance) is substantially equal to twice the operating rate of the system (the PWM rate in embodiments in which the driver is a PWM). In the embodiment shown in Figure 7, the decimator is integral to the autocorrelation and crosscorrelation estimator 320.

[0034] Figure 8 shows an embodiment of the system of these teachings in which the oversampling (sigma delta in one embodiment) modulator 310 is used as a sampling component and load current and state estimation is also performed. In the embodiment shown in Figure 8, the decimator is integral to the autocorrelation and crosscorrelation estimator 320 and the other decimator (288, Fig. 4) is also built-in. The use of a high-speed bit stream for digital converter control applies not just to correlators but also to compensators as well, in particular, even a simple proportional-integral-derivative (PID) compensator, when implemented in the velocity form. In this compensator, the error signal is processed by the sum of a gain, a gain times the derivative, and a gain times the second derivative. The processing result is then integrated. The decimation function, in this embodiment, is achieved by the final integrator and the dynamics of the power supply itself.

[0035] In one embodiment, the method of these teachings includes sampling an output signal from a system and an input signal from the system at a first sampling rate, the first sampling rate being at least equal to a predetermined operating rate, obtaining, from the sampled output signal and the sampled input signal, values for a number of input parameters, the values for the input parameters being sampled at the first sampling rate, decimating the values for the input parameter in order to obtain values for a number of subsampled input parameters, the subsampled input parameters being sampled at a second sampling rate, the second sampling rate been slower than the first sampling rate, obtaining, from the subsampled input parameters, a model for the system, obtaining, from the model for the system, compensator parameters and providing the compensator parameters to an adaptive compensator.

[0036] In one instance of the above described embodiment of the method of the present teachings, the input parameters for generating a model of the system are the autocorrelation and crosscorrelation. In one instance, simplified hardware can be used to calculate the auto- and cross-correlation. In many instances, the compensator does not have to be updated at every cycle thus the calculations of the planned estimate can be done at a much lower sampling rate. This allows the high-speed part of the algorithm to be implemented in specialized hardware and the low-speed part of the algorithm to be implemented in a very low-cost general purpose microprocessor.

[0037] In one instance, the high-speed algorithm may use a delta-sigma modulator as the ADC conversion element, as disclosed hereinabove. This high-speed, small-bit-width, oversampled data conversion method allows for simpler hardware. The typical delta-sigma ADC decimator can be integrated into or replaced by the correlation filter. Thus the correlation filter hardware is simplified.

[0038] A number of conventional techniques (algorithms) can be utilized for the adaptive identification of unknown dynamic systems. One of these conventional techniques is the least mean squares (LMS) algorithm. This technique can be easily implemented. However, the LMS algorithm can be slow to converge. The LMS algorithm is effective for the power supply converter application because good initial guesses can be given, so the power supply behaves properly before any adaptation is achieved. In many applications, it is desirable to use an algorithm that can identify the dynamics of the power supply in a time short compared to the time over which the system changes. For example, it is desirable for the algorithm to determine the dynamics of the object being controlled, in this case, the power electronics associated with the power supply, during startup and before regulation began.

[0039] One conventional technique for fast plant identification is the Recursive Least Squares (RLS) algorithm. This technique provides for fast convergence, however, it can suffer from a high computational burden. Also, the normal formulation of the Recursive Least Squares algorithm also has some assumptions in it that limit very high-speed performance.

[0040] The foundation of the use of autocorrelation and crosscorrelation for system identification comes from the statistical solution which minimizes the mean square error of the Wiener filter is given in equation (1) where is the vector of estimated filter coefficients, Rxx is the correlation matrix of the input signal, and rxy is the cross-correlation vector of the input and output signals.



[0041] It is possible to numerically estimate the auto-correlation matrix and cross-corrclation vector from the observed input and output signals of the plant (system) being identified. These estimates may be used directly to compute the estimated weight vector. In one embodiment of the method of the present teachings, the computational load is separated into two segments, one which must be performed on every new data sample, and one which must be performed only when it is desired to update the weight estimates. By scheduling the weight updates at a slower rate than the data, the overall computational burden of the algorithm can be reduced. However, when computed, the weight estimates will still make use of all data to that point. Thus, this method sacrifices only the weight update rate and not the quality of the estimates. Equation (2) depicts numerical estimates of the auto-correlation matrix and cross-correlation vector. The individual terms can then be computed using equations (3) and (4). Additionally, these expressions can be rewritten in a recursive manner such that they are updated incrementally with each new data point as given by (5) and (6), where the second index is the discrete time offset.











[0042] The optimal weight vector based on the numerical estimates can then be expressed as given in equation (7). Since the auto-correlation matrix has a Toeplitz structure, there are only 2P recursive estimates necessary for computation of the entire expression. Additionally, this structure allows for the use of efficient matrix inversion techniques which utilize this symmetry to reduce the number of required computations.



[0043] In another embodiment, the method of these teachings includes sampling an output signal from a system and an input signal from the system, obtaining, from the sampled output signal and the sampled input signal, values for a predetermined finite number of rows and columns from an inverse matrix and a predetermined finite number for a row vector in a least-squares solution, and obtaining, from the values for the predetermined finite number of rows and columns from an inverse matrix and the predetermined finite number for a row vector in a least-squares solution, a model for the system. Once a model of the system is obtained, an adaptive control method can be implemented.

[0044] While not desiring to be bound by theory, one rationale motivating the above described embodiment is given hereinbelow. The result of equation (7) above provides the same answer as the batch least squares solution in the case of infinite data. However, for any finite interval of data, the above result and the batch least squares result will differ. This can be seen directly by computing the batch mode least squares solution given by (8) over the same time interval [0,N-1], which results in the least squares weight vector given in equation (9). Comparison of the solutions in equation (7) and (9) indicate differences in the auto-correlation matrices. In fact, the least squares auto-correlation matrix of equation (9) is not necessarily Toeplitz for any finite data set. For a stationary input signal, and as the numerical sum approaches the true statistical value, that the two solutions (Equations (7) and (9)) are identical. However, in many applications, it is desirable to have the best possible estimate given a short finite interval of data. The estimate given by (9) minimizes the sum squared error over any finite data set.





[0045] The individual terms in equation (9) can, in one embodiment, be represented by iterative solutions and the computational load can be split as discussed above such that the weight vector solution is performed at a sub-sampled rate. However, there are now

unique entries in the auto-correlation matrix. It is not necessary to compute each of these separately since elements along each diagonal are simply a delayed version of first element of the diagonal as given by equation (10).



[0046] Given the above described property, and the symmetry of the matrix, only P running estimates need to be computed and the appropriate past values of each estimate must be stored. The auto-correlation matrix can then be computed directly from these values. Hereinbelow, this embodiment of the method of the present teachings is also referred as Iterative Least Squares since it exactly implements the least square solution over any finite interval of data.

[0047] The above described embodiment is generally applicable for both the FIR and IIR filters. In the instance where the above described embodiment is applied to IIR filters, the input matrix includes a mixture of both the past and current inputs as well as the past outputs.

[0048] It should be noted that the embodiments described hereinabove can be applied to a wide range of controlled components besides switching power supplies.

[0049] In order to better illustrate the systems and methods of the present teachings, results and details of several exemplary embodiments are presented hereinbelow. It should be noted that the methods and systems of these teachings are not limited to (or limited by) the illustrative embodiment presented hereinbelow.

[0050] In one illustrative embodiment, the Iterative Least Squares solution of equation (9) is compared with the method of equation (7) which uses the conventional numerical correlation estimates. In one instance, the two embodiments are simulated in a noise-free environment., The system to be identified is a 5 tap FIR filter with coefficients given in (9) and fed with a uniformly distributed random number sequence.



[0051] FIGURE 9 depicts, for the instance of a noiseless environment, two of the filter coefficients versus time. The dashed horizontal lines are for the Iterative Least Squares solution while the solid jagged lines for the conventional numerical correlation estimate method. In the noiseless case simulated, the Iterative Least Squares solution is correct after 5 time steps (the FIR filter length) while the solution based on the numerically calculated correlations takes more iterations to converge. Figure 11 depicts the sum squared error between the actual and estimated weight vector at each time step for the noiseless case. In Figure 11, the solid line represents the conventional correlation estimates and the dashed or lower line is for the Iterative Least Squares solution

[0052] In one instance, the Iterative Least Squares solution conventional numerical correlation estimate solution and are simulated for the case where the observations (filter outputs) are corrupted by uniformly-distributed random noise with an amplitude of 10% of the primary excitation signal. Results of the simulation for two of the filter coefficients versus time are shown in Figure 10. FIGURE 12 depicts the above described error signals for the instance where 10% white measurement noise is present.

[0053] In another illustrative embodiment, a controller was designed by assuming a second-order compensator and solving for the closed-loop transfer function. The denominator of the closed-loop transfer function was then equated to the desired transfer function. (The polynomial equation obtained by equating the desired transfer function to the denominator of the closed-loop transfer function is known as the Diophantine equation. See for example, Kelly, A., Rinne, K, Control of dc-dc converters by direct pole placement and adaptive feedforward gain adjustment, Twentieth Annual IEEE Applied Power Electronics Conference and Exposition, 2005; APEC 2005, Volume 3, Date: 6-10 March 2005, Pages: 1970 -1975.)

[0054] If the "plant" transfer function, obtained by identification, is given by a ratio of two polynomials, n(q)/d(q), where each polynomial is a polynomial in the z transform variable, and the second order deadbeat controller is also represented by a ratio of two polynomials, b(q)/a(q), the closed loop transfer function can be written as

the denominator of the above expression is of the form of a polynomial in z. Equating that desired characteristic polynomial to the denominator of the above equation provides a so-called Diophantine equation. In solving the Diophantine equation, it is the object to solve for the controller polynomials, giving a desired characteristic polynomial and the polynomials describing the plant. It is therefore possible to design a controller by direct pole placement. An exemplary embodiment is given herein below.

b(q)/a(q) = 2nd order deadbeat compensator

n(q)/d(q) = plant transfer function



[0055] Diophantine Equation:



[0056] Diophantine equation coefficients :



[0057] Matrix form of equation coefficients :



[0058] Multiplication of above expression for verification:



n1 :=0.37304 n2 :=0.24709 n3 :=-0.0014363
  d2 :=-1.8495 d3 :=0.87438


[0059] Diophantine equation matrix coefficients:



[0060] Diophantine solution:



[0061] Pole Locations (Closed loop and of compensator):

Closed Loop Poles:Compensator Poles:
 r :=polyroots (v)



r2 :=polyroots (v2)




[0062] Figure 13 depicts the transient response of a deadbeat controller, designed utilizing the methods of these teachings, to a step change in load current from 10A to 13A. In Figure 13, the upper trace is output voltage and the lower trace is the commanded duty cycle.

[0063] Figure 14 depicts the response of the deadbeat controller, designed utilizing the methods of these teachings, to a large load current step from 5A to 25A where the control signal saturates at full duty cycle. (In Figure 14, the upper curve reflects output voltage, and the lower curve reflects commanded duty cycle).

[0064] As described herein above, the state estimator provides a model of the system to be controlled with an extra input that is used to drive the error between the measurable variables and their corresponding quantities in the model. If the states are observable, it is possible to design a gain for this extra input that results in a stable estimator. However, zero error between the estimated states and the real on measurable states is not always obtained, particularly in the presence of unknown disturbances. This is a limitation when applying state space estimation in general and in particular when applying state space estimation for power supplies. Several embodiments of the state estimator used in these teachings are disclosed hereinbelow.

[0065] Switching power supplies can be modeled by a set of differential equations with time as the independent variable. As an example, the differential equations used to model a buck power supply are as follows:

Where: u is the input vector

x is the state vector

y is the output vector

Ae, Be, Ce, De are gain matrices



[0066] Assuming that PWM (while in one stationary mode) is modeled as a constant gain equal to the supply voltage, the state vector is the capacitor voltage and conductor current, the input is the duty cycle command, and the output is the output voltage, than the gain matrices are as follows:

Where: v is the capacitor voltage

i is the inductor current

Vdd is the supply voltage

duty is the PWM duty cycle

L is the inductor inductance

C is the capacitor capacitance

Rc is the parasitic resistance of the capacitor

R1 is the parasitic resistance of the inductor.



[0067] The Ae, Be, Cc, and De gain matrices can be readily identified from the equations above. The additional input for state estimation is the error between them measured values and their modeled values this error is multiplied by the gain L. and added to the state vector. In this embodiment, the first equation is modified as:

Where: Xest is the estimated states

yest is the estimate output

Le is the estimator gain matrix

Le can be calculator by conventional techniques.

[0068] Figure 15 is a graphical schematic presentation of a state space estimator. The results of a exemplary simulation using the buck converter equations can be seen in figure 16. It can be seen that there is a DC error in the inductor current. Also, the load current is not determined.

[0069] Another embodiment of the estimator of these teachings uses a different structure for the state space estimator where instead of using duty cycle as input, it attributes the error between the modeled output voltage and the measured output voltage primarily to an unknown load current. Using load current instead of the duty cycle input as input, the DC error is significantly reduced. Also, proportional integral control is used to drive the average DC error between the output voltage and the modeled output voltage. Figure 17a is a graphical schematic representation of another estimator structure embodiment of these teachings,

[0070] Referring to Figure 17a, there are two inputs duty cycle 410 and output voltage 420. The PWM or pulse with modulator is modeled as a constant gain 430 and its output is driven into a model of the switching power supply 460. The difference between the estimated output voltage and the measured output voltage, the output of substractor 440, is used to drive a compensator 450, which in the embodiment shown, is a proportional integral compensator. The compensator output is an estimate 480 of the load current disturbance and is fed into the load current input of the switching power supply model 460. The feedback action insures that the required estimated load disturbance is generated by the compensator. The switching power supply model 460 also provides an estimate of the state 470 and an estimate of the output voltage 490.

[0071] Figure 17b shows the discrete time model of a typical buck converter, one instance of the switching power supply model 460. The first integrator 510 is used to model the inductor current and the second integrator 520 is used to model the capacitor voltage. The inductor current and load current are summed to generate the capacitor current. Note that the sign of the inductor current is chosen to match that of a typical load. That means an increase in inductor current causes a reduction in the output voltage. The loops containing Rind and Reap model the losses and output voltage effects of the parasitic resistances.

[0072] Figure 18 is a graphical representations of results obtained, in an exemplary instance, using the embodiment of Figures 17a, 17b. Note that the output voltage is nearly accurately estimated, the inductor current substantially contains the expected DC information and the load current is estimated.

[0073] Figure 19 shows an additional embodiment, in which the duty cycle is taken as arising from a digital compensator and the discrete time converter model, low-pass filter and estimator compensator are implemented digitally. In this embodiment, the digital sampling rate is chosen to be a multiple of the PWM rate. The combination of that comparator and low-resolution DAC provides a noisy over sampled version of the estimation error.

[0074] Analog components are the comparator and DAC. As an alternative, a sigma delta modulator or other over-sampling modulator could be used. In this approach, the low-pass filter is the plant model and no additional large filtering delays are incurred. Figure 19 is an exemplary embodiment of an over-sampling state estimators that use sigma delta or other over-sampling modulator, such as the comparator and DAC, in the error feedback path. Such an embodiment has application to control systems other than control of switching power supplies and can result in a smaller delay than other embodiments.

[0075] Figure 20 is a graphical representations of results obtained, for an exemplary instance, using the embodiment of Figure 19.

[0076] It should be noted that, although the above description of the teachings utilized buck converters as an exemplary embodiment, these teachings are not limited to that embodiment. Also, although the above description of the teachings utilized sigma delta modulator as an exemplary embodiment of over-sampling modulators, other over-sampling modulators are within the scope of this invention. (See, for example, but not limited to, the over-sampling modulators disclosed in U.S. Application Publication No. 2XXX/XXXXXX, corresponding to U.S. Patent Application 11/550,893, entitled Systems and Methods for Digital Control.)

[0077] Although the present teachings have been described with respect to various embodiments, it should be realized these teachings are also capable of a wide variety of further and other embodiments within the scope of the appended claims.


Claims

1. A digital controller comprising:

a sampling component (220) configured to sample an output signal from a system (210) and an input signal from the system (210) at a first sampling rate, said first sampling rate being at least equal to a predetermined operating rate;

an input parameter obtaining component (230) configured to receive the output signal and the input signal sampled at the first sampling rate and to obtain values for a plurality of input parameters, said values for said plurality of input parameters being sampled at said first sampling rate;

a decimator component (240) configured to receive said values for said plurality of input parameters sampled at the first sampling rate and to provide subsampled values for said plurality of input parameters, said subsampled values being sampled at a second sampling rate that is slower than said first sampling rate;

an adaptive plant estimator component (250) configured to receive said subsampled values of said plurality of input parameters and to obtain a model of the system (210) based on said subsampled values; said model reflecting variations in the system (210);

a compensator design component (260) configured to receive the model of the system (210) and to provide compensator parameters based on said model; wherein said compensator parameters reflect changes in the system (210); values of said compensator parameters being sampled at the second sampling rate; and

a feedback compensator (270) configured to receive said values of said compensator parameters and to adjust said input signal to said system based on said compensator parameters; and further comprising:
a load current estimating component (280) configured to receive the output signal sampled at the first sampling rate and state estimation data from said adaptive plant estimator component (250) and to provide estimated load current data at the first sampling rate; and
another decimator component (288) configured to receive the estimated load current data at the first sampling rate and to provide the estimated load current data at a predetermined operating rate of the system (210) to said feedback compensator (270);
wherein the load current estimating component (280) comprises a state estimator (284) comprising:

a disturbance estimating component configured to provide an estimated disturbance from a difference between a sensed output and an estimated output; and

a power supply model component configured to receive the estimated disturbance as an input;

wherein said power supply model component also receives as input a duty cycle multiplied by a predetermined gain;

wherein said estimated disturbance is an estimated load current;

wherein said disturbance estimating component comprises a proportional integral component; and

wherein said power supply model provides an estimated state and said estimated output;

wherein said difference between said sensed output and said estimated output is obtained from an oversampling modulator (294);

said difference being an oversampled difference;

said oversampled difference being provided to said disturbance estimating component;

an output of said disturbance estimating component being an oversampled estimated disturbance; and

the state estimator (284) comprises:
a low pass filter for receiving said oversampled estimated disturbance and for providing said estimated disturbance.
 
2. The digital controller of claim 1, wherein said sampling component (220) is an analog to digital converter (ADC) for sampling the output signal and the input signal from the system (210) at the first sampling rate, said first sampling rate being at least equal to a predetermined operating rate of the system (210).
 
3. The digital controller of claim 1, wherein said input parameter obtaining component (230) provides a predetermined number of rows and columns from an inverse matrix and a row vector in a least-squares solution.
 
4. The digital controller of claim 1, wherein said sampling component (220) comprises an oversampling modulator (290); and wherein said first sampling rate is an oversampled rate.
 
5. The digital controller of claim 1, wherein the system (210) is a switching power supply and wherein a predetermined operating rate of the system is a PWM rate of said switching power supply.
 
6. The digital controller of claim 1, further comprising a disturbance injection component configured to provide a substantially small noise signal for detecting changes in the system (210).
 
7. A method for digital control of a system (210), the method comprising:

Sampling, using a sampling component (220), an output signal from a system and an input signal from the system at a first sampling rate, said first sampling rate being at least equal to a predetermined operating rate;

obtaining, from the sampled output signal and the sampled input signal, values for a plurality of input parameters using an input parameter obtaining component (230), said values for said plurality of input parameters being sampled at said first sampling rate;

decimating the values for the plurality of input parameters in order to obtain values for a plurality of subsampled input parameters using a first decimator component (240), the plurality of subsampled input parameters being sampled at a second sampling rate, the second sampling rate been slower than the first sampling rate;

obtaining, from the plurality of subsampled input parameters, a model for the system using an adaptive plant estimator component (250); said model reflecting variations in the system (210);

obtaining, from the model for the system, compensator parameters that reflect changes in the system using a compensator design component (260), wherein values of the compensator parameters are sampled at the second sampling rate; and

providing the compensator parameters to a feedback compensator (270); and adjusting said input signal to said system based on said compensator parameters;

obtaining, the output signal sampled at the first sampling rate and state estimation data, from said adaptive plant estimator component (250), and providing estimated load current data at the first sampling rate using a load current estimating component (280);

obtaining, the estimated load current data at the first sampling rate and providing the estimated load current data at a predetermined operating rate of the system (210) to said feedback compensator (270) using a second decimator component (288);

obtaining a difference between a sensed output from the system and an estimated output from an oversampling modulator (294);

oversampling the difference;

utilizing the oversampled difference in order to reduce a state estimation error, wherein the load current estimating component (280) comprises a state estimator (284) comprising:

a disturbance estimating component configured to provide an estimated disturbance from the difference between the sensed output and the estimated output; and

a power supply model component configured to receive the estimated disturbance as an input;

wherein said power supply model component also receives as input a duty cycle multiplied by a predetermined gain;

wherein said estimated disturbance is an estimated load current;

wherein said disturbance estimating component comprises a proportional integral component; and

wherein said power supply model provides an estimated state and said estimated output;
wherein said oversampled difference is provided to said disturbance estimating component, wherein an output of said disturbance estimating component being an oversampled estimated disturbance; and wherein the state estimator (284) comprises a low pass filter for receiving said oversampled estimated disturbance and for providing said estimated disturbance.
 
8. The method of claim 7, wherein the step of obtaining values for the plurality of input parameters comprises the step of obtaining estimates for an autocorrelation and a crosscorrelation.
 
9. The method of claim 7, wherein the step of obtaining values for the plurality of input parameters comprises the step of obtaining values for a predetermined number of rows and columns from an inverse matrix and a row vector in a least-squares solution.
 


Ansprüche

1. Digitale Steuerung, umfassend:

eine Abtastkomponente (220), die konfiguriert ist, um ein Ausgangssignal von einem System (210) und ein Eingangssignal von dem System (210) mit einer ersten Abtastgeschwindigkeit abzutasten, wobei die erste Abtastgeschwindigkeit zumindest einer vorgegebenen Betriebsgeschwindigkeit entspricht;

eine Eingangsparametererhaltungskomponente (230), die konfiguriert ist, um das Ausgangssignal und das Eingangssignal zu empfangen, die mit der ersten Abtastgeschwindigkeit abgetastet werden und um Werte für eine Vielzahl von Eingangsparametern zu erhalten, wobei die Werte für die Vielzahl von Eingangsparametern mit der ersten Abtastgeschwindigkeit abgetastet werden;

eine Dezimatorkomponente (240), die konfiguriert ist, um die Werte für die Vielzahl von Eingangsparametern zu empfangen, die mit der ersten Abtastgeschwindigkeit abgetastet werden und um unterabgetastete Werte für die Vielzahl von Eingangsparametern bereitzustellen, wobei die unterabgetasteten Werte mit einer zweiten Abtastgeschwindigkeit abgetastet werden, die langsamer ist als die erste Abtastgeschwindigkeit;

eine adaptive Anlagenschätzkomponente (250), die konfiguriert ist, um die unterabgetasteten Werte der Vielzahl von Eingangsparametern zu empfangen und um ein Modell des Systems (210) auf Grundlage der unterabgetasteten Werte zu erhalten; wobei das Modell Variationen in dem System (210) widerspiegelt; eine Kompensatordesignkomponente (260), die konfiguriert ist, um das Modell des Systems (210) zu empfangen und um Kompensatorparameter auf Grundlage des Modells bereitzustellen; wobei die Kompensatorparameter Veränderungen in dem System (210) widerspiegeln; wobei Werte der Kompensatorparameter mit der zweiten Abtastgeschwindigkeit abgetastet werden; und

einen Rückkopplungskompensator (270), der konfiguriert ist, um die Werte der Kompensatorparameter zu empfangen und um das Eingangssignal auf Grundlage der Kompensatorparameter an das System anzupassen; und ferner umfassend:

eine Laststromschätzkomponente (280), die konfiguriert ist, um das Ausgangssignal, das mit der ersten Abtastgeschwindigkeit abgetastet wird und Zustandsschätzdaten von der adaptiven Anlagenschätzkomponente (250) zu empfangen und um geschätzte Laststromdaten mit der ersten Abtastgeschwindigkeit bereitzustellen; und

eine andere Dezimatorkomponente (288), die konfiguriert ist, um die geschätzten Laststromdaten mit der ersten Abtastgeschwindigkeit zu empfangen und um die geschätzten Laststromdaten mit einer vorgegebenen Betriebsgeschwindigkeit des Systems (210) für den Rückkopplungskompensator (270) bereitzustellen;

wobei die Laststromschätzkomponente (280) eine Zustandsschätzvorrichtung (284) umfasst, umfassend:

eine Störungsschätzkomponente, die konfiguriert ist, um eine geschätzte Störung aus einer Differenz zwischen einer erfassten Ausgabe und einer geschätzten Ausgabe bereitzustellen; und

eine Stromversorgungsmodellkomponente, die konfiguriert ist, um die geschätzte Störung als eine Eingabe zu empfangen;

wobei die Stromversorgungsmodellkomponente außerdem als Eingabe einen Arbeitszyklus empfängt, der mit einer vorgegebenen Verstärkung multipliziert ist;

wobei die geschätzte Störung einem geschätzten Laststrom entspricht;

wobei die Störungsschätzkomponente eine proportionale integrale Komponente umfasst; und

wobei das Stromversorgungsmodell einen geschätzten Zustand und die geschätzte Ausgabe bereitstellt;

wobei die Differenz zwischen der erfassten Ausgabe und der geschätzten Ausgabe von einem Überabtastmodulator (294) erhalten wird;

wobei die Differenz einer überabgetasteten Differenz entspricht;

wobei die überabgetastete Differenz für die Störungsschätzkomponente bereitgestellt wird;

wobei eine Ausgabe der Störungsschätzkomponente einer überabgetasteten geschätzten Störung entspricht; und

wobei die Zustandsschätzvorrichtung (284) Folgendes umfasst:
einen Tiefpassfilter zum Empfangen der überabgetasteten geschätzten Störung und zum Bereitstellen der geschätzten Störung.


 
2. Digitale Steuerung nach Anspruch 1, wobei die Abtastkomponente (220) ein Analog-Digital-Wandler (ADC) zum Abtasten des Ausgangssignals und des Eingangssignals von dem System (210) mit der ersten Abtastgeschwindigkeit ist, wobei die erste Abtastgeschwindigkeit zumindest einer vorgegebenen Betriebsgeschwindigkeit des Systems (210) entspricht.
 
3. Digitale Steuerung nach Anspruch 1, wobei die Eingangsparametererhaltungskomponente (230) eine vorgegebene Anzahl von Zeilen und Spalten aus einer inversen Matrix und einen Zeilenvektor in einer Least-Squares-Lösung bereitstellt.
 
4. Digitale Steuerung nach Anspruch 1, wobei die Abtastkomponente (220) einen Überabtastmodulator (290) umfasst; und wobei die erste Abtastgeschwindigkeit eine überabgetastete Geschwindigkeit ist.
 
5. Digitale Steuerung nach Anspruch 1, wobei das System (210) ein Schaltnetzteil ist und wobei eine vorgegebene Betriebsgeschwindigkeit des Systems eine PWM-Geschwindigkeit des Schaltnetzteils ist.
 
6. Digitale Steuerung nach Anspruch 1, ferner umfassend eine Störungsinjektionskomponente, die konfiguriert ist, um ein im Wesentlichen kleines Rauschsignal zum Detektieren von Veränderungen in dem System (210) bereitzustellen.
 
7. Verfahren zur digitalen Steuerung eines Systems (210), wobei das Verfahren Folgendes umfasst:

Abtasten eines Ausgangssignal von einem System und eines Eingangssignals von dem System mit einer ersten Abtastgeschwindigkeit unter Verwendung einer Abtastkomponente (220), wobei die erste Abtastgeschwindigkeit zumindest einer vorgegebenen Betriebsgeschwindigkeit entspricht;

Erhalten von Werten für eine Vielzahl von Eingangsparametern unter Verwendung einer Eingangsparametererhaltungskomponente (230) aus dem abgetasteten Ausgangssignal und dem abgetasteten Eingangssignal, wobei die Werte für die Vielzahl von Eingangsparametern mit der ersten Abtastgeschwindigkeit abgetastet werden;

Dezimieren der Werte für die Vielzahl von Eingangsparametern, um Werte für eine Vielzahl von unterabgetasteten Eingangsparametern unter Verwendung einer ersten Dezimatorkomponente (240) zu erhalten, wobei die Vielzahl von unterabgetasteten Eingangsparametern mit einer zweiten Abtastgeschwindigkeit abgetastet wird, wobei die zweite Abtastgeschwindigkeit langsamer ist als die erste Abtastgeschwindigkeit;

Erhalten eines Modells für das System aus der Vielzahl von unterabgetasteten Eingangsparametern unter Verwendung einer adaptiven Anlagenschätzkomponente (250); wobei das Modell Variationen in dem System (210) widerspiegelt;

Erhalten von Kompensatorparametern, die Veränderungen in dem System widerspiegeln, unter Verwendung einer Kompensatordesignkomponente (260) aus dem Modell für das System, wobei Werte der Kompensatorparameter mit der zweiten Abtastgeschwindigkeit abgetastet werden; und

Bereitstellen der Kompensatorparameter für einen Rückkopplungskompensator (270); und Anpassen des Eingangssignals an das System auf Grundlage der Kompensatorparameter;

Erhalten des Ausgangssignals, das mit der ersten Abtastgeschwindigkeit abgetastet wird und von Zustandsschätzdaten von der adaptiven Anlagenschätzkomponente (250) und Bereitstellen geschätzter Laststromdaten mit der ersten Abtastgeschwindigkeit unter Verwendung einer Laststromschätzkomponente (280);

Erhalten der geschätzten Laststromdaten mit der ersten Abtastgeschwindigkeit und Bereitstellen der geschätzten Laststromdaten mit einer vorgegebenen Betriebsgeschwindigkeit des Systems (210) für den Rückkopplungskompensator (270) unter Verwendung einer zweiten Dezimatorkomponente (288);

Erhalten einer Differenz zwischen einer erfassten Ausgabe von dem System und einer geschätzten Ausgabe von einem Überabtastmodulator (294); Überabtasten der Differenz;

Verwenden der überabgetasteten Differenz, um einen Zustandsschätzfehler zu verringern, wobei die Laststromschätzkomponente (280) eine Zustandsschätzvorrichtung (284) umfasst, umfassend:

eine Störungsschätzkomponente, die konfiguriert ist, um eine geschätzte Störung aus der Differenz zwischen der erfassten Ausgabe und der geschätzten Ausgabe bereitzustellen; und

eine Stromversorgungsmodellkomponente, die konfiguriert ist, um die geschätzte Störung als eine Eingabe zu empfangen;

wobei die Stromversorgungsmodellkomponente außerdem als Eingabe einen Arbeitszyklus empfängt, der mit einer vorgegebenen Verstärkung multipliziert ist;

wobei die geschätzte Störung einem geschätzten Laststrom entspricht;

wobei die Störungsschätzkomponente eine proportionale integrale Komponente umfasst; und

wobei das Stromversorgungsmodell einen geschätzten Zustand und die geschätzte Ausgabe bereitstellt;

wobei die überabgetastete Differenz für die Störungsschätzkomponente bereitgestellt wird, wobei eine Ausgabe der Störungsschätzkomponente einer überabgetasteten geschätzten Störung entspricht; und wobei die Zustandsschätzvorrichtung (284) einen Tiefpassfilter zum Empfangen der überabgetasteten geschätzten Störung und zum Bereitstellen der geschätzten Störung umfasst.


 
8. Verfahren nach Anspruch 7, wobei der Schritt des Erhaltens von Werten für die Vielzahl von Eingangsparametern den Schritt des Erhaltens von Schätzungen für eine Autokorrelation und eine Kreuzkorrelation umfasst.
 
9. Verfahren nach Anspruch 7, wobei der Schritt des Erhaltens von Werten für die Vielzahl von Eingangsparametern den Schritt des Erhaltens von Werten für eine vorgegebene Anzahl von Zeilen und Spalten aus einer inversen Matrix und einen Zeilenvektor in einer Least-Squares-Lösung umfasst.
 


Revendications

1. Dispositif de commande numérique comprenant :

un composant d'échantillonnage (220) configuré pour échantillonner un signal de sortie provenant d'un système (210) et un signal d'entrée provenant du système (210) à un premier taux d'échantillonnage, ledit premier taux d'échantillonnage étant au moins égal à un taux de fonctionnement prédéterminé ;

un composant d'obtention de paramètres d'entrée (230) configuré pour recevoir le signal de sortie et le signal d'entrée échantillonnés au premier taux d'échantillonnage et pour obtenir des valeurs pour une pluralité de paramètres d'entrée, lesdites valeurs pour ladite pluralité de paramètres d'entrée étant échantillonnées audit premier taux d'échantillonnage ;

un composant de décimateur (240) configuré pour recevoir lesdites valeurs pour ladite pluralité de paramètres d'entrée échantillonnés au premier taux d'échantillonnage et pour fournir des valeurs sous-échantillonnées pour ladite pluralité de paramètres d'entrée, lesdites valeurs sous-échantillonnées étant échantillonnées à un second taux d'échantillonnage qui est plus lent que ledit premier taux d'échantillonnage ;

un composant d'estimateur d'installation adaptatif (250) configuré pour recevoir lesdites valeurs sous-échantillonnées de ladite pluralité de paramètres d'entrée et pour obtenir un modèle du système (210), sur la base desdites valeurs sous-échantillonnées ; ledit modèle reflétant des variations dans le système (210) ;

un composant de conception de compensateur (260) configuré pour recevoir le modèle du système (210) et pour fournir des paramètres de compensateur sur la base dudit modèle ; dans lequel lesdits paramètres de compensateur reflètent des changements dans le système (210) ; des valeurs desdits paramètres de compensateur étant échantillonnées au second taux d'échantillonnage ; et

un compensateur de rétroaction (270) configuré pour recevoir lesdites valeurs desdits paramètres de compensateur et pour ajuster ledit signal d'entrée vers ledit système sur la base desdits paramètres de compensateur ; et comprenant en outre :

un composant d'estimation de courant de charge (280) configuré pour recevoir le signal de sortie échantillonné au premier taux d'échantillonnage et des données d'estimation d'état à partir dudit composant d'estimateur d'installation adaptatif (250) et pour fournir des données de courant de charge estimé au premier taux d'échantillonnage ; et

un autre composant de décimateur (288) configuré pour recevoir les données de courant de charge estimé au premier taux d'échantillonnage et pour fournir les données de courant de charge estimé à un taux de fonctionnement prédéterminé du système (210) audit compensateur de rétroaction (270) ;

dans lequel le composant d'estimation de courant de charge (280) comprend un estimateur d'état (284) comprenant :

un composant d'estimation de perturbation configuré pour fournir une perturbation estimée à partir d'une différence entre une sortie détectée et une sortie estimée ; et

un composant de modèle d'alimentation électrique configuré pour recevoir la perturbation estimée comme une entrée ;

dans lequel ledit composant de modèle d'alimentation électrique reçoit également comme entrée un rapport cyclique multiplié par un gain prédéterminé ;

dans lequel ladite perturbation estimée est un courant de charge estimé ;

dans lequel ledit composant d'estimation de perturbation comprend un composant intégral proportionnel ; et

dans lequel ledit modèle d'alimentation électrique fournit un état estimé et ladite sortie estimée ;

dans lequel ladite différence entre ladite sortie détectée et ladite sortie estimée est obtenue à partir d'un modulateur de sur-échantillonnage (294) ;

ladite différence étant une différence sur-échantillonnée ;

ladite différence sur-échantillonnée étant fournie audit composant d'estimation de perturbation ;

une sortie dudit composant d'estimation de perturbation étant une perturbation estimée sur-échantillonnée ; et

l'estimateur d'état (284) comprend :
un filtre passe-bas pour recevoir ladite perturbation estimée sur-échantillonnée et pour fournir ladite perturbation estimée.


 
2. Dispositif de commande numérique selon la revendication 1, dans lequel ledit composant d'échantillonnage (220) est un convertisseur analogique-numérique (ADC) pour échantillonner le signal de sortie et le signal d'entrée provenant du système (210) au premier taux d'échantillonnage, ledit premier taux d'échantillonnage étant au moins égal à un taux de fonctionnement prédéterminé du système (210).
 
3. Dispositif de commande numérique selon la revendication 1, dans lequel ledit composant d'obtention de paramètres d'entrée (230) fournit un nombre prédéterminé de lignes et de colonnes à partir d'une matrice inverse et d'un vecteur de ligne dans une solution des moindres carrés.
 
4. Dispositif de commande numérique selon la revendication 1, dans lequel ledit composant d'échantillonnage (220) comprend un modulateur de sur-échantillonnage (290) ; et dans lequel ledit premier taux d'échantillonnage est un taux sur-échantillonné.
 
5. Dispositif de commande numérique selon la revendication 1, dans lequel le système (210) est une alimentation électrique à commutation et dans lequel un taux de fonctionnement prédéterminé du système est un taux de PWM de ladite alimentation électrique à commutation.
 
6. Dispositif de commande numérique selon la revendication 1, comprenant en outre un composant d'injection de perturbation configuré pour fournir un signal de bruit essentiellement petit pour détecter des changements dans le système (210).
 
7. Procédé de commande numérique d'un système (210), le procédé comprenant :

l'échantillonnage, en utilisant un composant d'échantillonnage (220), d'un signal de sortie provenant d'un système et un signal d'entrée provenant du système à un premier taux d'échantillonnage, ledit premier taux d'échantillonnage étant au moins égal à un taux de fonctionnement prédéterminé ;

l'obtention, à partir du signal de sortie échantillonné et du signal d'entrée échantillonné, des valeurs pour une pluralité de paramètres d'entrée en utilisant un composant d'obtention de paramètres d'entrée (230), lesdites valeurs pour ladite pluralité de paramètres d'entrée étant échantillonnées audit premier taux d'échantillonnage ;

la décimation des valeurs pour la pluralité de paramètres d'entrée de manière à obtenir des valeurs pour une pluralité de paramètres d'entrée sous-échantillonnés en utilisant un premier composant de décimateur (240), la pluralité de paramètres d'entrée sous-échantillonnés étant échantillonnés à un second taux d'échantillonnage, le second taux d'échantillonnage étant plus lent que le premier taux d'échantillonnage ;

l'obtention, à partir de la pluralité de paramètres d'entrée sous-échantillonnés, d'un modèle pour le système en utilisant un composant d'estimateur d'installation adaptatif (250) ; ledit modèle reflétant des variations dans le système (210) ; l'obtention, à partir du modèle pour le système, de paramètres de compensateur qui reflètent des changements dans le système en utilisant un composant de conception de compensateur (260), dans lequel des valeurs des paramètres de compensateur sont échantillonnées au second taux d'échantillonnage ; et

la fourniture des paramètres de compensateur à un compensateur de rétroaction (270) ; et l'ajustement dudit signal d'entrée vers ledit système sur la base desdits paramètres de compensateur ;

l'obtention du signal de sortie échantillonné au premier taux d'échantillonnage et de données d'estimation d'état à partir dudit composant d'estimateur d'installation adaptatif (250) et la fourniture de données de courant de charge estimé au premier taux d'échantillonnage en utilisant un composant d'estimation de courant de charge (280) ;

l'obtention des données de courant de charge estimé au premier taux d'échantillonnage et la fourniture des données de courant de charge estimé à un taux de fonctionnement prédéterminé du système (210) audit compensateur de rétroaction (270) en utilisant un second composant de décimateur (288) ;

l'obtention d'une différence entre une sortie détectée provenant du système et une sortie estimée provenant d'un modulateur de sur-échantillonnage (294) ;

le sur-échantillonnage de la différence ;

l'utilisation de la différence sur-échantillonnée de manière à réduire une erreur d'estimation d'état, dans lequel le composant d'estimation de courant de charge (280) comprend un estimateur d'état (284) comprenant :

un composant d'estimation de perturbation configuré pour fournir une perturbation estimée à partir de la différence entre la sortie détectée et la sortie estimée ; et

un composant de modèle d'alimentation électrique configuré pour recevoir la perturbation estimée comme une entrée ;

dans lequel ledit composant de modèle d'alimentation électrique reçoit également comme entrée un rapport cyclique multiplié par un gain prédéterminé ;

dans lequel ladite perturbation estimée est un courant de charge estimé ;

dans lequel ledit composant d'estimation de perturbation comprend un composant intégral proportionnel ; et

dans lequel ledit modèle d'alimentation électrique fournit un état estimé et ladite sortie estimée ;

dans lequel ladite différence sur-échantillonnée est fournie audit composant d'estimation de perturbation, dans lequel une sortie dudit composant d'estimation de perturbation étant une perturbation estimée sur-échantillonnée ; et dans lequel l'estimateur d'état (284) comprend un filtre passe-bas pour recevoir ladite perturbation estimée sur-échantillonnée et pour fournir ladite perturbation estimée.


 
8. Procédé selon la revendication 7, dans lequel l'étape d'obtention de valeurs pour la pluralité de paramètres d'entrée comprend l'étape d'obtention d'estimations pour une auto-corrélation et une corrélation croisée.
 
9. Procédé selon la revendication 7, dans lequel l'étape d'obtention de valeurs pour la pluralité de paramètres d'entrée comprend l'étape d'obtention de valeurs pour un nombre prédéterminé de lignes et de colonnes à partir d'une matrice inverse et d'un vecteur de ligne dans une solution des moindres carrés.
 




Drawing




































































Cited references

REFERENCES CITED IN THE DESCRIPTION



This list of references cited by the applicant is for the reader's convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard.

Patent documents cited in the description




Non-patent literature cited in the description