CN113641096B - Self-adaptive reconfigurable proportional-integral-derivative controller based on BP neural network - Google Patents
Self-adaptive reconfigurable proportional-integral-derivative controller based on BP neural network Download PDFInfo
- Publication number
- CN113641096B CN113641096B CN202110877611.4A CN202110877611A CN113641096B CN 113641096 B CN113641096 B CN 113641096B CN 202110877611 A CN202110877611 A CN 202110877611A CN 113641096 B CN113641096 B CN 113641096B
- Authority
- CN
- China
- Prior art keywords
- control
- neural network
- voltage
- pid
- switching
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 65
- 230000001052 transient effect Effects 0.000 claims abstract description 30
- 230000008859 change Effects 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 8
- 238000013459 approach Methods 0.000 claims description 7
- 230000033228 biological regulation Effects 0.000 claims description 6
- 230000007423 decrease Effects 0.000 claims description 5
- 230000003044 adaptive effect Effects 0.000 claims description 4
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 230000004044 response Effects 0.000 description 4
- 230000006872 improvement Effects 0.000 description 3
- 238000000034 method Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000012938 design process Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000000087 stabilizing effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B11/00—Automatic controllers
- G05B11/01—Automatic controllers electric
- G05B11/36—Automatic controllers electric with provision for obtaining particular characteristics, e.g. proportional, integral, differential
- G05B11/42—Automatic controllers electric with provision for obtaining particular characteristics, e.g. proportional, integral, differential for obtaining a characteristic which is both proportional and time-dependent, e.g. P. I., P. I. D.
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02M—APPARATUS FOR CONVERSION BETWEEN AC AND AC, BETWEEN AC AND DC, OR BETWEEN DC AND DC, AND FOR USE WITH MAINS OR SIMILAR POWER SUPPLY SYSTEMS; CONVERSION OF DC OR AC INPUT POWER INTO SURGE OUTPUT POWER; CONTROL OR REGULATION THEREOF
- H02M1/00—Details of apparatus for conversion
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
The invention relates to a self-adaptive reconfigurable proportional-integral-derivative (P-I-D) controller based on a BP neural network, which overcomes the influence of integral terms and differential terms on control performance in the existing PID controller so as to improve the transient performance of a digital power supply. The controller is based on data driving and mainly comprises a BP neural network and a reconfigurable controller. Firstly, data under PI, PD, PID control modes are collected respectively, then the BP neural network is trained offline by using the three groups of data respectively, and three groups of different control parameters (weight and bias of the neural network) are generated, wherein each group of control parameters represents one control mode. When the digital power supply works normally, the neural network generates a digital duty ratio according to parameters such as input voltage, inductance current, error voltage and the like so as to adjust output voltage to be stabilized at reference voltage. The reconfigurable controller adaptively switches control parameters of the neural network according to the real-time state of the output voltage of the digital power supply, thereby realizing the reconfiguration of the control mode.
Description
Technical Field
The invention belongs to the field of power electronics, and relates to a self-adaptive reconfigurable proportional-integral-derivative (P-I-D) controller based on a BP neural network, which is particularly suitable for intelligently controlling a digital power supply.
Background
The structure of the digitally controlled DC-DC switching converter (hereinafter referred to as a digital power supply) is shown in fig. 7. Analog output voltage V at load end out (t) conversion to digital output V by ADC out [k]Then V is taken out [k]With reference voltage V ref [k]Error signal e [ k ] between]And feeding into a digital compensator. In the digital compensator, a specific digital control algorithm (such as PID algorithm) is used to calculate the digital duty cycle signal d [ k ]]The digital duty cycle signal d [ k ] is then passed through a digital pulse width modulator (DPWM: digital Pulse Width Modulation)]Converted into an analog duty ratio signal d (t), and finally the power stage switch S is driven by the Gate driver p And S is n To adjust the output voltage V out (t) stabilizing it at the reference voltage V ref 。
Aiming at the control algorithm of the digital power supply, the traditional PID control algorithm is based on a linear small signal model, adopts fixed gain and control coefficient, has the advantages of simple structure and easy realization, but has poorer transient performance of the power supply. The control algorithm combining nonlinear control and PID control, such as fuzzy PID control and neural network PID control, has better steady state and transient performance. However, the adaptivity of fuzzy PID control is limited by the number of fuzzy subsets, and while increasing the number of fuzzy subsets can improve the adaptivity of control, the hardware resources required are greatly increased. The neural network PID control is usually based on data driving, can meet the requirements of each working point in theory when training data are enough, and has strong adaptability. However, the neural network PID control is still a fixed structure based on PID control, and cannot overcome the influence of integral terms and differential terms always existing in the PID control on the transient response of the power supply.
The documents Improvement of Compensation Effect of Neural Network Prediction for Digitally Controlled DC-DC Converter,2015IEEE International Telecommunications Energy Conference and A Reference Modification Model Digitally Controlled DC-DC Converter for Improvement of Transient Response, IEEE Trans.on Power elec, vol.31, no.1, jan.2016, successively propose a neural network PID controller applied to a digital Power supply, the neural network adopts an on-line training mode, and the control coefficient (k) of the PID controller is adjusted in real time according to the state variable of the digital Power supply p ,k i ,k d ) And reference voltage V ref The transient performance of the digital power supply is effectively improved. However, the fixed structure of the neural network PID controller is limited, and the structure and control parameters of the controller cannot be realized to be reconfigurable, so that the influence of an integral term and a differential term on the transient response of the power supply is difficult to eliminate, and the further improvement of the transient performance is limited.
Disclosure of Invention
Technical problem to be solved
Aiming at the influence of integral terms and differential terms in the existing PID controller on control performance, the transient performance of the digital power supply is improved. The invention provides a self-adaptive reconfigurable P-I-D controller based on a BP neural network. The controller not only can realize the self-adaptive adjustment of the control coefficient, but also can realize the real-time reconstruction of the controller structure. The disturbance recovery time and overshoot of the digital power supply are effectively reduced, and the transient performance is remarkably improved.
Technical proposal
An adaptive reconfigurable proportional-integral-derivative controller based on a BP neural network is characterized by comprising the BP neural network, a logic operation unit and a reconfigurable control unit; the logic operation unit carries out delay operation and addition and subtraction operation on the output voltage to generate voltage errors of the current period, the previous period and the voltage error change rate of the current period; the reconfigurable control unit adaptively switches control parameters under PI, PD, PID control modes of the BP neural network according to the real-time change state of the output voltage so as to realize the reconfiguration of the control modes; the BP neural network generates a digital duty ratio according to the input voltage, the inductance current and the voltage error change rate so as to adjust the output voltage to be stable to the reference voltage.
The reconfigurable control unit comprises a lookup table LUT storing PI, PD and PID control parameters, a multiplexer and a timing control module; the time sequence control module generates a control signal of the multiplexer according to the state variable of the output voltage; the lookup table LUT stores PI, PD and PID control parameters of the neural network; the multiplexer switches control parameters of the BP neural network according to the control signals to realize reconstruction in a control mode.
The BP neural network adopts a neural network with a single hidden layer, the structure is 6-10-1, namely 6 input nodes, 10 hidden nodes and 1 output node are adopted, then three groups of data are used for training the neural network in an off-line training mode, three groups of weights and biases are obtained through a gradient descent method, control parameters of PI, PD, PID control modes are represented respectively, and the trained three groups of control parameters are stored in a lookup table for standby.
An adaptive method of a reconfigurable proportional-integral-derivative controller, characterized by: the controller state switching has 6 cases, the switching is performed among PI, PD and PID, and the specific state switching time sequence is designed as follows:
(1) Switching from PI control to PD control, indicating that the digital power source is entering a transient state from a steady state, the controller needs to switch to PD control to regulate the output voltage so that it quickly approaches the reference voltage; the trigger condition (1) at this time is
(2) Switching from PD control to PID control, indicating that PD control reaches a limit, and voltage error cannot be reduced further, the controller needs to switch from PD control to PID control to eliminate voltage error, and the triggering condition (2) at this time is
e o [k]＝e o [k-1]≠0
(3) Switching from PID control to PI control indicates that transient regulation of the digital power supply is complete, and the digital power supply enters a steady state, and the controller can maintain steady state output only by PI control, and the triggering condition (3) at this time is
e o [k]＝e o [k-1]＝0，
(4) Switching from PD control to PI control, indicating that the digital power supply will enter steady state from transient state, and at transient state, PD control has just completely eliminated the voltage error, the output voltage is stabilized at the reference voltage; the triggering condition (4) at this time is
e o [k]＝e o [k-1]＝0
(5) The switching from PI control to PID control also shows that the digital power supply enters a transient state from a steady state, and the switching from PI control to PD control is different from the switching from PI control in that the disturbance is small, the rapid adjustment can be completed only by PID control, the disturbance is large in the latter, the output voltage deviates from the reference voltage more, the controller needs to switch to PD control to enable the output voltage to rapidly approach the reference voltage, and then PID is adopted to eliminate the voltage error; the trigger condition (5) for switching from PI control to PID control is therefore
e oc [k]＜e oc [k-1]
(6) Switching from PID control to PD control, indicating that one disturbance adjustment of the digital power supply is incomplete and a new disturbance, namely superposition of the disturbances, is also achieved; originally under the regulation of PID control, the voltage error gradually decreases, suddenly gets new disturbance, and increases again, the controller needs to switch from the original PID control to PD control to quickly reduce the voltage error, so the triggering condition (6) is
{e o [k]＞e o [k-1]}||{e o [k]＝e o [k-1]≠0}。
Advantageous effects
Compared with the prior art, the self-adaptive reconfigurable P-I-D controller based on the neural network is data driven, and can meet the requirements of all working points. The neural network controller based on data driving not only has the self-adaption of control coefficients, but also can be adaptively switched among PI, PD, PID control modes according to the state of output voltage, so that the real-time reconfigurable of the proportional-integral-derivative controller structure is realized. The self-adaptive reconfigurable P-I-D controller based on the BP neural network reduces the influence of integral terms and differential terms in a PID controller on the transient performance of a power supply, and effectively improves the transient performance of a digital power supply.
Drawings
The drawings are only for purposes of illustrating particular embodiments and are not to be construed as limiting the invention, like reference numerals being used to refer to like parts throughout the several views.
FIG. 1 shows an adaptive reconfigurable P-I-D controller based on BP neural network;
FIG. 2 illustrates a reconfigurable mechanism of a controller according to the present invention;
FIG. 3 illustrates the working principle of the reconfigurable controller in the load disturbance;
FIG. 4 is a flow of acquisition of neural network parameters;
fig. 5 illustrates a neural network hardware circuit architecture employed by the present invention: (a) a normalization module of the neural network; (b) circuitry for inputting layers to hidden layers by the BP neural network; (c) circuitry underlying the layer to the output layer;
FIG. 6 is a comparison of digital power supply transient performance using the inventive technique with existing neural network-PID (NN-PID) techniques: (a) an activated state; (b) the load current suddenly changes from 1.2A to 0.2A; (c) the load current suddenly changes from 0.2A to 1.2A; (d) the input voltage suddenly changes from 5.0V to 6.0V; (e) the input voltage suddenly changes from 5.0V to 4.0V;
fig. 7 is a diagram of a digitally controlled DC-DC switching converter (simply referred to as a digital power supply).
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention. In addition, technical features of the embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
Referring to fig. 1, the self-adaptive reconfigurable P-I-D controller based on the BP neural network provided by the invention consists of the BP neural network, a logic operation unit and a reconfigurable control unit, wherein the reconfigurable control unit consists of a lookup table (LUT) for storing PI, PD and PID control parameters, a multiplexer and a timing control module. The input signal of the self-adaptive reconfigurable P-I-D controller based on BP neural network is { V } out [k],I ind [k],V in [k]}, wherein V out [k]Is the output voltage of the current period, I ind [k]For inductor current of current period, V in [k]For the input voltage of the current period, the output signal is a digital duty cycle d nn [k]. The logic operation unit outputs the voltage V out [k]Performing delay operation and addition and subtraction operation to generate { e } o [k],e oc [k],e o [k-1],e o [k-2]E, where e o [k]Is the voltage error of the current period, e o [k-1]And e o [k-2]The voltage errors of the previous period and the previous period are respectively, e oc [k]＝e o [k]-e o [k-1]Is the voltage error rate of the current cycle. The reconfigurable control unit adaptively switches according to the real-time change state of the output voltageThe control parameters of the neural network are exchanged to realize the reconfigurability of the control mode. Wherein the time sequence control module is used for controlling the output voltage according to the state variable { e }, of the output voltage o [k],e oc [k]The control signal ctrl of the multiplexer is generated. The look-up table stores three sets of weights and biases for the neural network, namely PI, PD and PID control parameters. The multiplexer switches the weight and bias of the BP neural network according to the control signal ctrl to realize the reconstruction in a control mode. BP neural network according to input voltage V in Inductor current I ind Voltage error information { e } o [k],e o [k-1],e o [k-2],e oc [k]Generating a digital duty cycle d nn [k]To regulate the output voltage. The specific design process of the time sequence control module is as follows:
1. and acquiring parameters of the neural network. The flow of acquiring parameters of the neural network is shown in FIG. 4, wherein data of three control modes (PI, PD, PID) are collected respectively, namely input information { V ] of the proposed controller in ,I ind ,e o [k],e o [k-1],e o [k-2],e oc [k]Output information { d } nn [k]}. In order to avoid the influence of different dimensionalities of the data on the control precision of the neural network, the acquired data is normalized. Input signal X m [k]The algorithm for (m=1, …, 6) normalization is as follows:
in the invention, a neural network with a single hidden layer is adopted, the structure is 6-10-1, namely 6 input nodes, 10 hidden nodes and 1 output node are adopted. And then training the neural network by using three groups of data in an off-line training mode through a gradient descent method to obtain three groups of weights and biases which respectively represent the control parameters of the PI, PD, PID three control modes, and storing the trained three groups of control parameters into a lookup table for standby. The built circuit is shown in fig. 5, wherein fig. 5a is a normalization module of the neural network, fig. 5b is a circuit from an input layer to an hidden layer of the BP neural network, and fig. 5c is a circuit from the hidden layer to an output layer.
2. The design of the reconfigurable controller. The reconfigurable controller adaptively switches control parameters of the neural network according to the state of the output voltage and is mainly used for selecting the control parameters of the neural network, so that the control mode can be reconfigured. As shown in fig. 1, the reconfigurable controller is composed of a look-up table (LUT), a multiplexer, and a timing control module. The reconfigurable control state is shown in fig. 2. In the starting stage of the power supply or in the corresponding initial stage of the transient state, as the output voltage deviates more from the reference voltage, the reconfigurable controller firstly selects the PD control parameter to realize PD control, so that the output voltage quickly approaches the reference voltage (the PD control parameter refers to the parameter trained by the neural network under the data of PD control, and the PI control parameter and the PID control parameter are the same). The conditions under which the controller is maintained in PD control are as follows:
{e o [k]≠e o [k-1]}&{e o [k]＞e PD [k]}&{e o [k-1]＞e PD [k]}， (4)
indicating that the voltage error is constantly changing in the PD control mode, this means that the PD control does not reach the limit and still can continue to reduce the voltage error. Wherein e o [k]E is the voltage error of the current period o [k-1]E is the voltage error of the previous period PD [k]Indicating voltage errors that cannot be eliminated when the PD control reaches the limit.
Because the PD control can not completely eliminate the voltage error, when the PD control can not continuously reduce the voltage error, the reconfigurable controller outputs PID control parameters, the PID control is realized to eliminate the voltage error, and the conditions of the controller maintained in the PID control are as follows:
{e o [k]≠e o [k-1]}&{0＜e o [k]＜e PD [k]}&{0＜e o [k-1]＜e PD [k]}， (5)
again, this means that the voltage error is constantly changing, meaning that the voltage error has not been completely eliminated.
When the voltage error is eliminated, the output voltage is equal to the reference voltage, the controller is switched to PI control, and the controller enters a steady state to keep the output voltage stable at the reference voltage. And in a steady state, the neural network adopts PI control parameters to realize PI control, so that the output voltage is maintained to be stable at the reference voltage. The conditions under which the controller maintains PI control are:
{e oc [k]＝e oc [k-1]＝0}&{e o [k]＝0}， (6)
the change rate of the voltage error is always zero, and the voltage error is also zero, and the output voltage is stabilized at the reference voltage. Wherein eoc [ k ] represents the error rate of the current cycle, and eoc [ k-1] represents the error rate of the previous cycle.
The controller state switching has 6 cases, the switching is performed among PI, PD and PID, and the specific state switching time sequence is designed as follows:
(1) Switching from PI control to PD control, representing the digital power supply going from steady state to transient, the controller needs to switch to PD control to regulate the output voltage so that it quickly approaches the reference voltage. The trigger condition (1) at this time is
Wherein e oc [k]>e oc [k-1]Representing the speed of the output voltage deviating from the reference voltage increasingly faster; e, e oc [k]＝e oc [k-1]The voltage is equal to 0, and the output voltage deviates from the reference voltage at a positive uniform speed; e, e oc [k]＝e oc [k-1]Not equal to 0 and e o [k]＝e o [k-1]Not equal to 0 indicates that the output voltage deviates from the reference voltage and does not change.
(2) Switching from PD control to PID control, indicating that PD control reaches a limit, and voltage error cannot be reduced further, the controller needs to switch from PD control to PID control to eliminate voltage error, and the triggering condition (2) at this time is
e o [k]＝e o [k-1]≠0， (8)
Indicating that the voltage error does not continue to decrease, i.e. the PD control reaches the limit.
(3) Switching from PID control to PI control indicates that transient regulation of the digital power supply is complete, and the digital power supply enters a steady state, and the controller can maintain steady state output only by PI control, and the triggering condition (3) at this time is
e o [k]＝e o [k-1]＝0， (9)
That is, both the voltage error and the error rate are 0, indicating that the output voltage has stabilized at the reference voltage.
(4) Switching from PD control to PI control indicates that the digital power supply will go from transient to steady state and at transient, the PD control happens to completely eliminate the voltage error and the output voltage stabilizes at the reference voltage. The triggering condition (4) at this time is
e o [k]＝e o [k-1]＝0， (10)
The same trigger condition as the PID control is switched to PI control.
(5) The switching from PI control to PID control also indicates that the digital power supply enters a transient state from a steady state, unlike PI control to PD control, which has small disturbance and can be quickly adjusted by only PID control, which has large disturbance and large deviation of the output voltage from the reference voltage, the controller needs to switch to PD control to quickly approximate the output voltage to the reference voltage, and then uses PID to eliminate the voltage error. The trigger condition (5) for switching from PI control to PID control is therefore
e oc [k]＜e oc [k-1]， (11)
Indicating that the error rate gradually decreases, i.e. the output voltage deviates from the reference voltage more and more slowly.
(6) Switching from PID control to PD control indicates that one disturbance adjustment of the digital power supply is incomplete and a new disturbance, i.e., superposition of disturbances, has occurred. Originally under the regulation of PID control, the voltage error gradually decreases, suddenly gets new disturbance, and increases again, the controller needs to switch from the original PID control to PD control to quickly reduce the voltage error, so the triggering condition (6) is
{e o [k]＞e o [k-1]}||{e o [k]＝e o [k-1]≠0}， (12)
Wherein e o [k]>e o [k-1]Indicating that the voltage error is gradually increasing, i.e. the output voltage is deviating from the reference voltage; e, e o [k]＝e o [k-1]Not equal to 0 indicates that the output voltage deviates from the reference voltage and does not change any more.
Examples:
the working principle of the reconfigurable control and the timing control will be described here by taking load disturbance as an example. As shown in fig. 3, when the load current suddenly changes, the output voltage deviates from the reference voltage, and once the trigger condition (1) is satisfied, the neural network is switched from PI control to PD control so that the output voltage approaches the reference voltage as soon as possible, as shown in equation (1).
Wherein e oc [k]>e oc [k-1]The speed of the output voltage deviating from the reference voltage is increasing; e, e oc [k]＝e oc [k-1]The voltage is equal to 0, and the output voltage deviates from the reference voltage at a positive uniform speed; e, e oc [k]＝e oc [k-1]Not equal to 0 and e o [k]＝e o [k-1]Not equal to 0 indicates that the output voltage deviates from the reference voltage and does not change.
With the gradual reduction of the voltage error, when the voltage error meets the triggering condition (2), as shown in the formula (2), the PD control reaches the limit, and the voltage error cannot be continuously eliminated, and at the moment, the neural network is switched from the PD control to the PID control so as to eliminate the voltage error and reduce the disturbance recovery time.
e o [k]＝e o [k-1]≠0， (2)
When e [ k ] =e [ k-1] =0, i.e. the trigger condition (3) is satisfied, the steady state error indicating the output voltage is eliminated, and the neural network is switched from PID control to PI control to maintain steady state.
For Buck type digital power supply, figure 6 is a transient response curve comparison of the digital power supply in the starting state, load current jump + -1A and input voltage jump + -1V by adopting the technology of the invention and the existing neural network-PID (NN-PID) control technology respectively.
While the invention has been described with reference to certain preferred embodiments, it will be understood by those skilled in the art that various changes and substitutions of equivalents may be made without departing from the spirit and scope of the invention.
Claims (2)
1. An adaptive reconfigurable proportional-integral-derivative controller based on a BP neural network is characterized by comprising the BP neural network, a logic operation unit and a reconfigurable control unit; the logic operation unit carries out delay operation and addition and subtraction operation on the output voltage to generate voltage errors of the current period, the previous period and the voltage error change rate of the current period; the reconfigurable control unit adaptively switches control parameters under PI, PD, PID control modes of the BP neural network according to the real-time change state of the output voltage so as to realize the reconfiguration of the control modes; the BP neural network generates a digital duty ratio according to the input voltage, the inductance current and the voltage error change rate so as to adjust the output voltage to be stabilized at the reference voltage;
the reconfigurable control unit comprises a lookup table LUT storing PI, PD and PID control parameters, a multiplexer and a timing control module; the time sequence control module generates a control signal of the multiplexer according to the state variable of the output voltage; the lookup table LUT stores PI, PD and PID control parameters of the neural network; the multiplexer switches control parameters of the BP neural network according to the control signals to realize reconstruction in a control mode;
the state switching of the self-adaptive reconfigurable proportional-integral-derivative controller is 6 cases, the mutual switching among PI, PD and PID is realized, and the specific state switching time sequence is designed as follows:
(1) Switching from PI control to PD control, indicating that the digital power source is entering a transient state from a steady state, the controller needs to switch to PD control to regulate the output voltage so that it quickly approaches the reference voltage; the trigger condition (1) at this time is
(2) Switching from PD control to PID control, indicating that PD control reaches a limit, and voltage error cannot be reduced further, the controller needs to switch from PD control to PID control to eliminate voltage error, and the triggering condition (2) at this time is
e o [k]＝e o [k-1]≠0
(3) Switching from PID control to PI control indicates that transient regulation of the digital power supply is complete, and the digital power supply enters a steady state, and the controller can maintain steady state output only by PI control, and the triggering condition (3) at this time is
e o [k]＝e o [k-1]＝0，
(4) Switching from PD control to PI control, indicating that the digital power supply will enter steady state from transient state, and at transient state, PD control has just completely eliminated the voltage error, the output voltage is stabilized at the reference voltage; the triggering condition (4) at this time is
e o [k]＝e o [k-1]＝0
(5) The switching from PI control to PID control also shows that the digital power supply enters a transient state from a steady state, and the switching from PI control to PD control is different from the switching from PI control in that the disturbance is small, the rapid adjustment can be completed only by PID control, the disturbance is large in the latter, the output voltage deviates from the reference voltage more, the controller needs to switch to PD control to enable the output voltage to rapidly approach the reference voltage, and then PID is adopted to eliminate the voltage error; the trigger condition (5) for switching from PI control to PID control is therefore
e oc [k]＜e oc [k-1]
(6) Switching from PID control to PD control, indicating that one disturbance adjustment of the digital power supply is incomplete and a new disturbance, namely superposition of the disturbances, is also achieved; originally under the regulation of PID control, the voltage error gradually decreases, suddenly gets new disturbance, and increases again, the controller needs to switch from the original PID control to PD control to quickly reduce the voltage error, so the triggering condition (6) is
{e o [k]＞e o [k-1]}||{e o [k]＝e o [k-1]≠0}。
2. The self-adaptive reconfigurable proportional-integral-derivative controller based on the BP neural network according to claim 1, wherein the BP neural network adopts a neural network with a single hidden layer, the structure is 6-10-1, namely, 6 input nodes are 10 hidden nodes, 1 output node is 1, then three groups of data are used for training the neural network in an off-line training mode through a gradient descent method, three groups of weights and offsets are obtained, the three groups of control parameters respectively represent control parameters of PI, PD, PID control modes, and the trained three groups of control parameters are stored in a lookup table for standby.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110877611.4A CN113641096B (en) | 2021-08-01 | 2021-08-01 | Self-adaptive reconfigurable proportional-integral-derivative controller based on BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110877611.4A CN113641096B (en) | 2021-08-01 | 2021-08-01 | Self-adaptive reconfigurable proportional-integral-derivative controller based on BP neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113641096A CN113641096A (en) | 2021-11-12 |
| CN113641096B true CN113641096B (en) | 2024-02-02 |
Family
ID=78419263
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110877611.4A Active CN113641096B (en) | 2021-08-01 | 2021-08-01 | Self-adaptive reconfigurable proportional-integral-derivative controller based on BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113641096B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101917119A (en) * | 2010-08-23 | 2010-12-15 | 东南大学 | High transient response digital control system and method of switch power supply |
| CN108566088A (en) * | 2018-04-13 | 2018-09-21 | 杭州电子科技大学 | Two close cycles RBF neural sliding moding structure self-adaptation control method |
| CN108566086A (en) * | 2018-04-13 | 2018-09-21 | 杭州电子科技大学 | Two close cycles RBF neural sliding moding structure adaptive control system |
| CN108809167A (en) * | 2018-06-26 | 2018-11-13 | 长春工业大学 | A kind of BP neural network PID speed regulating control algorithms based on fuzzy control |
| CN111245313A (en) * | 2020-03-16 | 2020-06-05 | 爱科赛智能科技(浙江)有限公司 | Brushless direct current motor PID speed regulation method based on genetic neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10181791B2 (en) * | 2017-04-14 | 2019-01-15 | Allegro Microsystems, Llc | Converter digital control circuit with adaptive compensation |
-
2021
- 2021-08-01 CN CN202110877611.4A patent/CN113641096B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101917119A (en) * | 2010-08-23 | 2010-12-15 | 东南大学 | High transient response digital control system and method of switch power supply |
| CN108566088A (en) * | 2018-04-13 | 2018-09-21 | 杭州电子科技大学 | Two close cycles RBF neural sliding moding structure self-adaptation control method |
| CN108566086A (en) * | 2018-04-13 | 2018-09-21 | 杭州电子科技大学 | Two close cycles RBF neural sliding moding structure adaptive control system |
| CN108809167A (en) * | 2018-06-26 | 2018-11-13 | 长春工业大学 | A kind of BP neural network PID speed regulating control algorithms based on fuzzy control |
| CN111245313A (en) * | 2020-03-16 | 2020-06-05 | 爱科赛智能科技(浙江)有限公司 | Brushless direct current motor PID speed regulation method based on genetic neural network |
Non-Patent Citations (2)
| Title |
|---|
| 基于BP网络的PID整定控制;陈玉芳;;兵工自动化(第05期);全文 * |
| 模糊RBF太阳电池最大功率点跟踪算法;袁娇;李涛;李文全;贺乔卿;李宸宇;马昊翔;;传感器与微***(第01期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113641096A (en) | 2021-11-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110048606B (en) | DC-DC boost converter dynamic sliding mode voltage control method based on interval two-type self-adaptive fuzzy neural network | |
| US7545130B2 (en) | Non-linear controller for switching power supply | |
| Samosir et al. | Dynamic evolution control for synchronous buck DC–DC converter: Theory, model and simulation | |
| Leng et al. | A simple model predictive control for Buck converter operating in CCM | |
| Jagannathan | Control of a class of nonlinear discrete-time systems using multilayer neural networks | |
| Dhali et al. | Pwm-based sliding mode controller for dc-dc boost converter | |
| CN111371322B (en) | Boost type converter control method and system based on finite time convergence observer | |
| Yuan et al. | Design of a single-input fuzzy PID controller based on genetic optimization scheme for DC-DC buck converter | |
| CN113641096B (en) | Self-adaptive reconfigurable proportional-integral-derivative controller based on BP neural network | |
| Gireesh et al. | High frequency SEPIC converter with PWM integral sliding mode control | |
| Mude et al. | Adaptive control schemes for DC-DC buck converter | |
| Dhivya et al. | Neural network controller for boost converter | |
| Liu et al. | A backpropagation neural network controller trained using PID for digitally-controlled DC-DC switching converters | |
| Mamat et al. | Analysis of SEPIC-Boost converter using several PID feedback tuning methods for renewable energy applications | |
| CN111810701B (en) | Method for realizing variable dynamic characteristics of electromagnetic valve at each stage by switching multiple voltage sources | |
| CN112636598A (en) | Control method, device and system of synchronous BUCK circuit and electronic device | |
| Shi et al. | A reinforcement learning-based online-training AI controller for DC-DC switching converters | |
| KR102193264B1 (en) | DC-DC Converter with Intelligent Controller | |
| CN115037157A (en) | ISOS system output voltage balance control method based on model prediction | |
| Rasheed | Performance of the adaptive sliding mode control scheme for output voltage control of the DC/DC buck converter system | |
| Goudarzian et al. | Voltage regulation of a negative output luo converter using a pd-pi type sliding mode current controller | |
| CN112415890A (en) | fuzzy-PID digital voltage compensator for simultaneously correcting error factor and PID control coefficient | |
| Anitha et al. | Design of linear and nonlinear controller for DC-DC boost converter with right-half plane zero | |
| Suthar et al. | Proximate time-optimal control of flying-capacitor multi-level converters using a fixed frequency pid framework | |
| Thirumeni et al. | Performance analysis of PI and SMC controlled zeta converter |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |