Dynamic internal model separation and gain adjustment-based repetitive control disturbance rejection method

Abstract

The invention discloses a repetitive control disturbance denial method based on dynamic internal model separation and gain adjustment, and the method comprises the following steps: building a system model of an initial control system, which comprises an initial controller; an internal model separation channel and an RC gain adjustment channel are added in a traditional repetitive controller of an initial controller, a target controller based on variable gain repetitive control aperiodic disturbance rejection is obtained, and the target controller comprises a repetitive control aperiodic disturbance rejection controller internal model, a separation threshold judgment device, an internal model separator and an RC gain adjuster; judging whether the internal model is a dynamic separation threshold value or not according to the separation threshold value judging device, and further judging whether the internal model of the repetitive control aperiodic disturbance rejection controller is separated or not; according to the invention, the internal model separator and the RC gain regulator are used for sensitively and quickly separating the internal model under the condition of short-time disturbance, and gain regulation is carried out, so that the repeated control performance is prevented from being reduced.

Description

Technical Field

[0001] The present invention relates to the field of industrial control technology, and more particularly to a repeated control disturbance rejection method based on dynamic internal model separation and gain adjustment. Background Art

[0002] Repetitive control (RC) has the ability to accurately track periodic signals of specific frequencies and has a strong ability to suppress disturbances at these frequencies. The excellent performance of RC under periodic signals makes it widely used in power, motor, motion control and other fields. The ideal internal model is a key link in the construction of RC. The ideal internal model can iterate the signal cycle by cycle and gradually optimize the control performance under the action of negative feedback; however, the time delay of the ideal internal model will cause the lag of RC regulation. This lag will lead to poor control performance and even damage the equipment under the action of short-term non-periodic disturbances. Summary of the invention

[0003] The purpose of the present invention is to provide a repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment, which can quickly and sensitively separate the internal model under short-term disturbance conditions through an internal model separator and an RC gain adjuster, and perform gain adjustment to prevent the repetitive control performance from decreasing.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment comprises the following steps:

[0006] Establishing the system model of the initial control system, including the initial controller;

[0007] An internal model separation channel and an RC gain adjustment channel are added to the traditional repetitive controller of the initial controller to obtain a target controller based on variable gain repetitive control non-periodic disturbance rejection, wherein the target controller includes an internal model of a repetitive control non-periodic disturbance rejection controller, a separation threshold determiner, an internal model separator and an RC gain adjuster;

[0008] According to the determination of the separation threshold determiner, when the error signal of the target controller is less than the dynamic separation threshold, the repetitive control non-periodic disturbance rejection controller internal model maintains a non-separation state, and the RC gain regulator does not perform gain adjustment;

[0009] When the error signal is greater than the dynamic separation threshold, the internal model separator is used to separate the internal model of the repetitive control non-periodic disturbance rejection controller, and the output gain of the target controller is gain-adjusted through the output gain of the RC gain adjuster.

[0010] Furthermore, the inner mold separator is specifically:

[0011] A first multiplier is connected between the repetitive control gain output signal of the initial controller and the internal model input of the repetitive control non-periodic disturbance rejection controller.

[0012] Furthermore, the construction method of the RC gain regulator is specifically as follows:

[0013] The reference signal I of the initial controller ref With the feedback signal I g Take the absolute value respectively and create a branch;

[0014] The reference signal I ref The absolute value of the feedback signal I g The absolute value of the scalar error e va , specifically:

[0015] e va =|I ref |-|I g |

[0016] The scalar error e va connected to the RC gain module via a second multiplier;

[0017] The separation threshold determiner is connected to the second multiplier through the NOT module.

[0018] Furthermore, the construction method of the separation threshold judger is specifically as follows:

[0019] Create a branch for the error signal e of the traditional repetitive controller and take the absolute value to obtain the absolute value of the error e a =|e|;

[0020] The absolute value of the error e a Enter the half-cycle delay and get the signal branch e after the half-cycle delay alast , specifically:

[0021] e alast =e a × -N / 2

[0022] The signal e alast Multiply the sensitivity coefficient k to get the dynamic separation threshold gate, which is:

[0023] gate=e alase ×k

[0024] The comparison module is used to compare the absolute value of the error e a Compared with the dynamic separation threshold gate; when the absolute value of the error e aWhen the dynamic separation threshold gate is smaller than the dynamic separation threshold gate, the comparison module outputs 1, and the internal model signal e is repeatedly controlled by the first multiplier. k No changes occur, and the inner mold does not separate;

[0025] When the absolute value of the error e a When it is greater than the dynamic separation threshold gate, the comparison module outputs 0, and the internal model signal e of the non-periodic disturbance rejection controller is repeatedly controlled by the first multiplier. k becomes 0 to realize the internal model separation of the repetitive control non-periodic disturbance rejection controller; at this time, the signal output by the comparison module is 1 after being inverted by the NOT module, and the RC gain module performs gain adjustment on the output of the target controller.

[0026] Furthermore, the RC gain regulator outputs a gain to perform gain regulation on the output signal of the target controller, specifically: the output signal of the target controller is gain-adjusted through a gain regulation function, and is regulated in combination with a negative feedback mechanism.

[0027] Furthermore, the scalar error is used to unify the direction of the error, specifically:

[0028] When the scalar error is negative, it means that the control output is too large;

[0029] When the scalar error is positive, it means that the control output is too small.

[0030] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0031] 1. Dynamic internal model separation technology based on error delay signal, through the determined delay time, uses the information of the corresponding beats of the adjacent periods of the error signal to decide whether the RC internal model is separated, and uses the error signal itself to determine the separation threshold. The dynamic separation threshold is adaptive and can sensitively and quickly separate the RC internal model under various conditions, preventing short-term non-periodic disturbances from entering the RC internal model, thereby preventing the "deterioration stage" of the RC control effect.

[0032] 2. The repeated control variable gain technology based on the negative feedback principle can change the output gain through the RC gain regulator when a short-term disturbance occurs, thereby indirectly forming a closed loop for the control action of RC, preventing the repeated control from entering the "out-of-control stage", and effectively improving the problems exposed by RC when facing short-term disturbances. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0034] The following is a further description of the disturbance rejection method of the present invention using repeated control based on dynamic internal model separation and gain adjustment in conjunction with the accompanying drawings;

[0035] Figure 1 It is a schematic diagram of the overall flow of the repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment provided by the present invention;

[0036] Figure 2 It is a control system of a traditional repetitive controller in a repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment provided by the present invention;

[0037] Figure 3 It is a control system of a target controller for repeated control non-periodic disturbance rejection based on variable gain in a repeated control disturbance rejection method based on dynamic internal model separation and gain adjustment provided by the present invention;

[0038] Figure 4 It is a waveform diagram of the grid voltage drop within half a cycle of a traditional repetitive controller in the repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment provided by the present invention;

[0039] Figure 5 It is a grid-connected current waveform diagram of a conventional repetitive controller in a method for repetitive control disturbance rejection based on dynamic internal model separation and gain adjustment provided by the present invention when the controller is in an out-of-control stage and a deterioration stage;

[0040] Figure 6 It is a periodic variation characteristic diagram of the error signal in the repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment provided by the present invention;

[0041] Figure 7 The absolute value of the error and the signal branch after half-cycle delay in the repeated control disturbance rejection method based on dynamic internal model separation and gain adjustment provided by the present invention are alast Relationship diagram in the time domain;

[0042] Figure 8 It is a slope diagram for negative feedback regulation in the repetitive control disturbance rejection method based on dynamic internal model separation and gain regulation provided by the present invention;

[0043] Fig. 9It is a diagram showing the effect of adding only an internal model separation channel to the repeated control disturbance rejection method based on dynamic internal model separation and gain adjustment provided by the present invention;

[0044] Fig.10 It is an effect diagram of simultaneously adding an internal model separation channel and an RC gain adjustment channel to the repeated control disturbance rejection method based on dynamic internal model separation and gain adjustment provided by the present invention. DETAILED DESCRIPTION

[0045] The specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0046] In order to better understand the purpose, structure and function of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings.

[0047] RC's unique tracking and suppression capabilities for periodic signals come from the ideal internal model link composed of the delay link and the positive feedback loop. However, RC has obvious disadvantages for short-term non-periodic disturbances; the existence of the delay link makes the control effect of RC hysteresis; when short-term disturbances occur, the control effect of RC will not be adjusted immediately, causing the system to be in a semi-out-of-control state (out-of-control stage) for a short period of time. When RC begins to produce a regulating effect, the short-term disturbance has disappeared, and the control effect of RC will deteriorate the control effect (deterioration stage).

[0048] Take the grid-connected inverter as an example to analyze the problem. In the grid-connected inverter, the grid voltage is a periodic external disturbance; the repetitive control system can effectively suppress the disturbance caused by the grid voltage, thereby obtaining high-quality grid-connected current. However, in some special cases, the grid voltage may drop, which is caused by power equipment failure or instantaneous load increase. The grid voltage drop is a short-term non-periodic disturbance, such as Figure 4 As shown, the grid voltage drops to 20% of its amplitude within half a cycle, and then the grid recovers.

[0049] like Figure 5 As shown in the figure, the solid line is the grid-connected current, and the dotted line is the reference value of the grid-connected current. When a short-term disturbance occurs, due to the hysteresis of the control action, the RC still outputs a large voltage in the half cycle of the voltage drop to offset the grid voltage output. However, since the grid is in a drop state at this time, the output of the RC is too large, the current is out of control, and may even cause overcurrent; after a cycle, the RC internal model uses the newly introduced error to start regulating. The RC begins to reduce the output, but the drop has recovered at this time. This reduction leads to insufficient resistance to disturbances, deterioration of the grid-connected current, and backflow.

[0050] Therefore, Figure 1As shown, the present invention provides a repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment, comprising the following steps:

[0051] like Figure 2 As shown, a system model of an initial control system is established, including an initial controller;

[0052] It should be noted that: Figure 2 I ref is the reference signal, I g is the control system feedback signal; e is the error signal; Q(z) is the internal model filter; z -N is the delay link; k r is the repetitive control gain; S(z) is the phase compensator; y rc For repeated control output.

[0053] like Figure 3 As shown, an internal model separation channel and an RC gain adjustment channel are added to the traditional repetitive controller of the initial controller to obtain a target controller based on variable gain repetitive control non-periodic disturbance rejection, wherein the target controller includes an internal model of a repetitive control non-periodic disturbance rejection controller, a separation threshold determiner, an internal model separator and an RC gain adjuster;

[0054] According to the determination of the separation threshold determiner, when the error signal of the target controller is less than the dynamic separation threshold, the internal model of the repetitive control non-periodic disturbance rejection controller is not separated, and the RC gain regulator does not work;

[0055] When the error signal is greater than the dynamic separation threshold, the internal model separator is used to separate the internal model of the repetitive control non-periodic disturbance rejection controller, and the output gain of the target controller is changed by the RC gain regulator to perform gain adjustment.

[0056] The inner mold separator is specifically:

[0057] A first multiplier is provided between the repetitive control gain of the initial controller and the inner model of the repetitive control non-periodic disturbance rejection controller.

[0058] The RC gain regulator is specifically:

[0059] The reference signal I of the initial controller ref With the feedback signal I g Take the absolute value respectively and create a branch;

[0060] The reference signal I ref The absolute value of the feedback signal I g The absolute value of the scalar error e va , specifically:

[0061] e va =|I ref |-|I g |

[0062] The scalar error e va connected to the RC gain module via a second multiplier;

[0063] The separation threshold determiner is connected to the second multiplier through the NOT module.

[0064] The separation threshold judger is specifically:

[0065] Create a branch for the error signal e of the traditional repetitive controller and take the absolute value to obtain the absolute value of the error e a =|e|;

[0066] It should be noted that: Figure 6 As shown, since the control system of the traditional repetitive controller works under the action of a periodic signal under normal working conditions, the error signal e also presents a periodic variation characteristic; Figure 6 The dashed line is the error e, and the solid line is the absolute value of the error e a , the absolute value of the error e a It can be roughly regarded as a signal of halving the cycle.

[0067] The absolute value of the error e a Enter the half-cycle delay and get the signal branch e after the half-cycle delay alast , specifically:

[0068] e alast =e a × -N / 2

[0069] It should be noted that: Figure 7 As shown, e a With e alast In the time domain, the difference is half a power grid cycle. alast Multiply the sensitivity coefficient k to get the dynamic separation threshold gate, which is:

[0070] gate=e alase ×k

[0071] The absolute value of the error e is calculated by the comparison module. a Compared with the dynamic separation threshold gate; when the absolute value of the error e a When the dynamic separation threshold gate is smaller than the dynamic separation threshold gate, the comparison module outputs 1, and the internal model signal e is repeatedly controlled by the first multiplier. k No changes occur, and the inner mold does not separate;

[0072] When the absolute value of the error ea When it is greater than the dynamic separation threshold gate, the comparison module outputs 0, and the internal model signal e of the non-periodic disturbance rejection controller is repeatedly controlled by the first multiplier. k becomes 0 to realize the internal model separation of the repetitive control non-periodic disturbance rejection controller; at this time, the signal output by the comparison module is 1 after being inverted by the NOT module, and the RC gain regulator performs gain adjustment on the output of the target controller.

[0073] It should be noted that the sensitivity coefficient determines the ease of RC internal model separation. When the sensitivity coefficient is large, the dynamic separation threshold is large, the separation condition is difficult to trigger, the control system is insensitive to short-term disturbances, and the RC internal model can receive most of the error information. Only when the error produces a large mutation can the separation condition be met. When the sensitivity coefficient is small, the system is more sensitive to short-term disturbances, and smaller short-term disturbances can also cause internal model separation. This dynamic threshold scheme can quickly and sensitively determine the occurrence of short-term disturbances. Since the size of short-term disturbances is a relative concept, if the judgment threshold is designed as a fixed value, only short-term disturbances of fixed size can be distinguished. The dynamic threshold essentially introduces the concept of relative size into the algorithm, obtains the current error and compares it with the corresponding position error of the previous cycle, and determines whether the disturbance occurring at this time is a short-term disturbance from the perspective of relative size.

[0074] The structure after adding internal model separation and variable gain is as follows Figure 6 , e a Compared with gate, if e a Not greater than gate, the comparison module outputs 1, and the internal model signal e is repeatedly controlled k After passing through the first multiplier at the repetitive control internal model input, the signal does not change (the repetitive control internal model input signal e k The output of the comparison module is then inverted by the NOT module to obtain 0. va The signal is disconnected after the second multiplier of the signal branch (e va Multiplying by 0 equals 0), the variable gain algorithm becomes invalid, and the system is now a traditional repetitive control.

[0075] If e a is greater than gate, which is equivalent to the system judging a short-term burst disturbance, the comparison module outputs 0, and after repeatedly controlling the first multiplier at the internal model input end, the signal e k becomes 0, the signal no longer enters the repeated control internal model, and the internal model is separated k The signal cannot enter the repetitive control internal model. At the same time, the signal output by the comparison module is inverted by the NOT module and becomes 1. va Branch open (e va Multiply by 1 to get e va ), variable gain is activated.

[0076] Therefore, through internal model separation, short-term disturbances can be prevented from affecting the RC internal model, thereby avoiding the effect of these disturbances being amplified in the subsequent control process, effectively curbing the deterioration of the control effect.

[0077] The RC gain module performs gain adjustment on the output of the target controller, specifically:

[0078] Through the gain adjustment function, the output gain of the target controller is negatively feedback adjusted.

[0079] It should be noted that: the RC output gain is adjusted by negative feedback through the gain adjustment function. Figure 8 The linear gain function shown is taken as an example: y=0.1x+1, wherein the slope of the linear gain adjustment function can be adjusted. The larger the slope, the stronger the negative feedback effect, but the more likely it is to be unstable. In this embodiment, the slope is taken as 0.1.

[0080] When the scalar error is near zero, the output gain of the repetitive control is close to 1, indicating that the RC output y rc The control effect can be maintained well, and the output remains unchanged after the gain adjustment function out =y rc ; When the scalar error is positive, it means that the output gain is insufficient. The algorithm increases the output gain according to the gain function to make y out >y rc , otherwise reduce the output gain y out <y rc .

[0081] The linear gain adjustment function essentially forms a closed-loop feedback between the output of the repetitive control and the system output in real time, and adjusts the RC output gain in real time through the negative feedback formed by the scalar error and the gain adjustment function. This negative feedback is temporarily constructed during the internal model separation. Compared with the feedback of the repetitive control system, this feedback is real-time and will not be affected by the RC internal model delay link, eliminating the hysteresis of the feedback.

[0082] In order to unify the direction of the error, the scalar error does not need to distinguish the direction of the current, taking the current as an example. As long as the scalar error is negative, it means that the control output is too large, and if it is positive, it means that the control output is too small.

[0083] The present invention also takes the grid-connected inverter repetitive control system as an example to conduct simulation verification of the technical solution in the present invention:

[0084] like Fig. 9 As shown, the present invention uses the inner mold separation technology to successfully curb the decline in control performance. Fig. 9It can be seen that by timely separating the inner model, the short-term disturbance is effectively blocked and does not penetrate into the RC inner model. Therefore, the control effect is successfully prevented from further deteriorating in the subsequent process.

[0085] The present invention effectively suppresses short-term non-periodic disturbances by simultaneously adopting internal mode separation and RC output gain adjustment technology. Fig.10 It can be clearly seen that the repetitive control strategy based on dynamic internal model separation and internal model output gain adjustment can significantly eliminate the impact of short-term non-periodic disturbances on the control system.

[0086] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment, characterized in that: The following steps are involved: Establishing the system model of the initial control system, including the initial controller; An internal model separation channel and an RC gain adjustment channel are added to the traditional repetitive controller of the initial controller to obtain a target controller based on variable gain repetitive control non-periodic disturbance rejection, wherein the target controller includes an internal model of a repetitive control non-periodic disturbance rejection controller, a separation threshold determiner, an internal model separator and an RC gain adjuster; According to the determination of the separation threshold determiner, when the error signal of the target controller is less than the dynamic separation threshold, the repetitive control non-periodic disturbance rejection controller internal model maintains a non-separation state, and the RC gain regulator does not perform gain adjustment; When the error signal is greater than the dynamic separation threshold, the internal model separator is used to separate the internal model of the repetitive control non-periodic disturbance rejection controller, and the output gain is changed by the RC gain regulator to perform gain adjustment on the output signal of the target controller.

2. The repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment according to claim 1 is characterized in that: The inner mold separator is specifically: A first multiplier is connected between the repetitive control gain output signal of the initial controller and the internal model input of the repetitive control non-periodic disturbance rejection controller.

3. The repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment according to claim 2 is characterized in that: The construction method of the RC gain regulator is specifically as follows: The reference signal I of the initial controller ref With the feedback signal I g Take the absolute value respectively and create a branch; The reference signal I ref The absolute value of the feedback signal I g The absolute value of the scalar error e va , specifically: yes va =|I ref |-|I g | The scalar error e va connected to the RC gain module via a second multiplier; The separation threshold determiner is connected to the second multiplier through the NOT module.

4. The repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment according to claim 3 is characterized in that: The construction method of the separation threshold judger is specifically as follows: Create a branch for the error signal e of the traditional repetitive controller and take the absolute value to obtain the absolute value of the error e a =|e|; The absolute value of the error e a Enter the half-cycle delay and get the signal branch e after the half-cycle delay alast , specifically: And alast =and a ×z -N / 2 The signal e alast Multiply the sensitivity coefficient k to get the dynamic separation threshold gate, which is: gate=and alase ×k The comparison module is used to compare the absolute value of the error e a Compared with the dynamic separation threshold gate; when the absolute value of the error e a When the dynamic separation threshold gate is smaller than the dynamic separation threshold gate, the comparison module outputs 1, and the internal model signal e is repeatedly controlled by the first multiplier. k No changes occur, and the inner mold does not separate; When the absolute value of the error e a When it is greater than the dynamic separation threshold gate, the comparison module outputs 0, and the internal model signal e of the non-periodic disturbance rejection controller is repeatedly controlled by the first multiplier. k becomes 0 to realize the internal model separation of the repetitive control non-periodic disturbance rejection controller; at this time, the signal output by the comparison module is 1 after being inverted by the NOT module, and the RC gain module performs gain adjustment on the output of the target controller.

5. The repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment according to claim 1 is characterized in that: The RC gain regulator outputs a gain to perform gain regulation on the output signal of the target controller, specifically: the output signal of the target controller is gain-adjusted through a gain regulation function, and is regulated in combination with a negative feedback mechanism.

6. The repetitive control disturbance rejection method based on dynamic internal model separation and gain adjustment according to claim 3 is characterized in that: The scalar error is used to unify the direction of the error, specifically: When the scalar error is negative, it means that the control output is too large; When the scalar error is positive, it means that the control output is too small.

Images