Boost converter control method and device and electronic equipment

Abstract

The invention discloses a control method and device of a boost converter and electronic equipment. The method comprises the following steps: acquiring a target moment output voltage value and a target moment inductance current value of a boost converter model; determining a state vector of the target moment according to the target moment output voltage value and the target moment inductance current value, the state vector at the target moment is at least used for representing a deviation value between an output voltage value at the target moment of the boost converter and a voltage reference value and a deviation value between an inductive current value at the target moment and an inductive current reference value, and the inductive current reference value is generated through a PID control loop; a pre-trained learning model is adopted to analyze the state vector, an action vector is obtained, the action vector is used for being converted into a switching signal, and the switching signal is used for controlling on-off of a switching tube in the boost converter so as to control the output voltage value of the boost converter to be equal to the voltage reference value.

Description

Technical Field

[0001] The present application relates to the field of microgrid control technology, and in particular to a control method, device and electronic device for a boost converter. Background Art

[0002] Under the influence of global environmental pollution and energy crisis, distributed power generation based on renewable energy has been widely used. In recent years, with the widespread application of distributed power systems such as solar and wind power, more and more power electronic cascade systems have been applied to power supply systems.

[0003] Boost DC-DC converters are widely used in photovoltaic systems, energy storage systems, electric vehicles, and other fields because they transfer power from a low-voltage DC power source to a high-voltage electronic load. The most traditional control method uses a proportional-integral-derivative (PID) controller based on pulse-width modulation (PWM). Specifically, the PID parameters are determined based on an established system average model. However, standard average models often ignore the effects of switching frequency, resulting in poor control accuracy.

[0004] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0005] The embodiments of the present application provide a control method, device and electronic device for a boost converter, so as to at least solve the technical problem in the related art of using a PID control method to control a boost converter, resulting in low control accuracy.

[0006] According to one aspect of an embodiment of the present application, a control method for a boost converter is provided, comprising: collecting an output voltage value and an inductor current value of a boost converter model at a target moment; determining a state vector at the target moment based on the output voltage value and the inductor current value at the target moment, wherein the state vector at the target moment is used to at least represent a deviation value between the output voltage value of the boost converter at the target moment and a voltage reference value and a deviation value between the inductor current value and the inductor current reference value at the target moment, and the inductor current reference value is generated by a proportional-integral-differential controller (PID) control loop; analyzing the state vector using a pre-trained learning model to obtain an action vector, wherein the action vector is used to convert into a switching signal, and the switching signal is used to control the on and off of a switch tube in the boost converter to control the output voltage value of the boost converter to be equal to the voltage reference value.

[0007] Optionally, determining the state vector based on the output voltage value at the target moment and the inductor current value at the target moment includes: determining a reference value of the inductor current based on the output voltage value at the target moment and the voltage reference value; determining a voltage tracking error based on the output voltage value at the target moment and the voltage reference value, and determining the inductor current tracking error based on the inductor current value at the target moment and the reference value of the inductor current; respectively obtaining the inductor current value at the target moment and an integral value of the inductor current tracking error; and determining the state vector at the target moment based on the output voltage value at the target moment, the voltage tracking error, the inductor current value at the target moment, and the integral value of the inductor current tracking error.

[0008] Optionally, a pre-trained learning model is used to analyze the state vector to obtain an action vector, including: creating an initial actor network, an initial critic network, a target actor network and a target critic network; collecting the state vectors of the boost converter model at multiple moments to form a state space, and collecting the action vectors corresponding to the state vectors at the multiple moments to form an action space; using the state vectors at multiple moments in the state space and the action vectors at multiple moments in the action space to train the initial actor network and the initial critic network to obtain the learning model; using the learning model to analyze the state vector to obtain the action vector, the action vector is used to represent the duty cycle, and the duty cycle represents the ratio of the conduction time of the switch tube in the boost converter to the switching period.

[0009] Optionally, the initial actor network and the initial critic network are trained using state vectors at multiple moments in the state space and action vectors at multiple moments in the action space to obtain the learning model, including: successively acquiring the state vector of each acquisition moment from the state space according to a preset step size; analyzing the state vector of each acquisition moment using the initial actor network to obtain the action vector corresponding to the state vector of each acquisition moment, and adding noise to the action vector corresponding to the state vector of each acquisition moment to obtain a first action vector; evaluating the first action vector using the initial critic network to obtain a reward corresponding to the first action vector; respectively determining the state vector of each acquisition moment, the first action vector, the reward corresponding to the first action vector, and the state vector of the next moment as experience tuples, and placing the experience tuples into a preset experience area, the state vector of the next moment representing the state vector after executing the action corresponding to the first action vector; and using the experience tuples in the preset experience area to update the model parameters of the initial actor network, the initial critic network, the target actor network, and the target critic network to obtain the learning model.

[0010] Optionally, the experience tuples in the preset experience area are used to update the model parameters of the initial critic network, including: extracting experience tuples from the preset experience area, and using the initial actor network to predict the experience tuples to obtain the action vector at the next moment corresponding to the first action vector; adding noise to the action vector at the next moment to obtain a second action vector; using the initial critic network to evaluate the second action vector to obtain an evaluation function value of the second action vector, and the evaluation function value is used to represent the cumulative reward value of the second action vector; determining the predicted evaluation function value of the first action vector based on the evaluation function value of the second action vector; determining the loss function value based on the predicted evaluation function value of the first action vector and the actual evaluation function value of the first action vector; and updating the model parameters of the initial critic network based on the loss function value.

[0011] Optionally, the initial critic network is used to evaluate the second action vector to obtain an evaluation function value of the second action vector, including: respectively using the first critic network and the second critic network in the initial critic network to evaluate the second action vector to obtain a first evaluation value and a second evaluation value; and determining the smaller value of the first evaluation value and the second evaluation value as the evaluation function value of the second action vector.

[0012] Optionally, the model parameters of the target reviewer network are updated using the experience tuples in the preset experience area, including: obtaining the updated model parameters of the initial reviewer network in each update step; and updating the updated model parameters to the model parameters of the target reviewer network using a soft update method.

[0013] According to another aspect of an embodiment of the present application, a non-volatile storage medium is further provided. The non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute any one of the control methods for a boost converter.

[0014] According to another aspect of an embodiment of the present application, an electronic device is provided, including: a processor; and a memory for storing processor-executable instructions, wherein the processor is configured to execute the instructions to implement any one of the control methods for a boost converter.

[0015] In an embodiment of the present application, a PID control method is used to generate a reference inductor current, and a predetermined learning model is used to obtain the output voltage. The PID control method and the model algorithm are combined to achieve the purpose of accurately controlling the output voltage value of the boost converter, thereby achieving the technical effect of improving the control accuracy, and further solving the technical problem of low control accuracy caused by using the PID control method to control the boost converter in the related art. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0017] Figure 1 is a hardware structure block diagram of a computer terminal for implementing a control method for a boost converter according to an embodiment of the present application;

[0018] Figure 2 is a flow chart of a control method of a boost converter according to an embodiment of the present application;

[0019] Figure 3 is a structural diagram of a boost converter according to an embodiment of the present application;

[0020] Figure 4 is a training flow chart of a learning model according to an embodiment of the present application;

[0021] Figure 5 is a flow chart of another method for controlling a boost converter according to an embodiment of the present application;

[0022] Figure 6 is a simulation schematic diagram of a control method for a boost converter according to an embodiment of the present application;

[0023] Figure 7 is a simulation schematic diagram of another control method for a boost converter according to an embodiment of the present application;

[0024] Figure 8 3 is a schematic structural diagram of an optional control device of a boost converter according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] In order to solve the problems existing in the related art, the embodiment of the present application provides a control method for a boost converter. Figure 1 In the computer terminal shown, the computer terminal is explained below.

[0028] The control method embodiment of the boost converter provided in the embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG1 shows a hardware structure block diagram of a computer terminal for implementing a control method for a boost converter. Figure 1 As shown, the computer terminal 10 may include one or more (illustrated by 102a, 102b, ..., 102n in the figure) processors (the processor may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission module 106 for communication functions connected via a wired and / or wireless network. In addition, it may also include: a display, a keyboard, a cursor control device, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, and a BUS bus. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0029] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuitry." The data processing circuitry may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuitry may be a single, independent processing module, or may be incorporated in whole or in part into any of the other components of the computer terminal 10. As described in the embodiments of the present application, the data processing circuitry serves as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0030] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the control method of the boost converter in the embodiment of the present application. The processor executes the software programs and modules stored in the memory 104 to perform various functional applications and data processing, thereby implementing the control method of the boost converter described above. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include memory remotely located relative to the processor, and these remote memories may be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0031] The transmission module 106 is configured to receive or transmit data via a network. A specific example of the aforementioned network may include a wireless network provided by the communications provider of the computer terminal 10. In one embodiment, the transmission module 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, the transmission module 106 may be a radio frequency (RF) module, which is configured to communicate with the Internet wirelessly.

[0032] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 .

[0033] It should be noted that, in some optional embodiments, the above Figure 1 The computer terminal shown may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of hardware elements and software elements. Figure 1This is merely one example of a particular embodiment and is intended to illustrate the types of components that may be present in the computer terminal described above.

[0034] In the above-mentioned operating environment, an embodiment of the present application provides an embodiment of a control method for a boost converter. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0035] Figure 2 is a control method of a boost converter according to an embodiment of the present application, such as Figure 2 As shown, the method includes the following steps:

[0036] Step S202 , collecting the output voltage value and the inductor current value of the boost converter model at the target moment;

[0037] Figure 3 A schematic diagram of the structure of a boost converter is shown in FIG. Figure 3 As shown, the boost converter includes an inductor, a load, a DC voltage source, a capacitor, and a resistor, and the power supply is connected to the load through the boost converter.

[0038] The boost converter model is shown below:

[0039]

[0040]

[0041] Where V in Represents the input voltage of the boost converter, V o represents the output voltage of the boost converter, L represents the inductor, C represents the capacitor, R represents the load resistance, P represents the load power, i L represents the inductor current, d represents the duty cycle of the switch tube, t represents the time, represents the rate of change of the inductor current, Indicates the rate of change of output voltage, P CPL Indicates the rated power of the load, i CPL Indicates the load current.

[0042] Step S204: determining a state vector at a target moment based on the output voltage value at the target moment and the inductor current value at the target moment, wherein the state vector at the target moment is used to represent at least a deviation between the output voltage value of the boost converter at the target moment and a voltage reference value, and a deviation between the inductor current value at the target moment and an inductor current reference value, wherein the inductor current reference value is generated by a proportional-integral-derivative (PID) control loop.

[0043] In step S204, the process of generating the inductor current reference value through the PID control loop is as follows: collecting the output voltage value and the inductor current value corresponding to the target time; performing PID control on the output voltage value and the voltage reference value to generate the inductor current reference value.

[0044] In step S206, a pre-trained learning model is used to analyze the state vector to obtain an action vector, wherein the action vector is used to convert into a switching signal, and the switching signal is used to control the on and off of the switch tube in the boost converter to control the output voltage value of the boost converter to be equal to the voltage reference value.

[0045] This boost converter control method utilizes a PID control method to generate a reference inductor current and uses a predetermined learning model to determine the output voltage. This method combines the PID control method with a model algorithm to precisely control the output voltage of the boost converter, thereby achieving the technical effect of improving control accuracy. This method addresses the technical issues of low control accuracy associated with using PID control methods for boost converters in related technologies. This is described in detail below.

[0046] It should be noted that the present application adopts a predetermined learning model, for example, a Twin Delayed Deep Deterministic Policy Gradient model to replace the traditional PID control method, and adopts the voltage control action generated by the learning model without being affected by external interference.

[0047] In some embodiments of the present application, the specific steps of determining the state vector based on the output voltage value at the target moment and the inductor current value at the target moment are as follows: determining a reference value of the inductor current based on the output voltage value at the target moment and the voltage reference value; determining a voltage tracking error based on the output voltage value at the target moment and the voltage reference value, and determining the inductor current tracking error based on the inductor current value at the target moment and the reference value of the inductor current; respectively obtaining the inductor current value at the target moment and the integral value of the inductor current tracking error; and determining the state vector at the target moment based on the output voltage value at the target moment, the voltage tracking error, the inductor current value at the target moment, and the integral value of the inductor current tracking error.

[0048] With the state vector S t ={V o (t),V o_e (t),i L (t),i L_e (t),∫i L_e} as an example, where Vo (t) represents the output voltage value at the target time, V o_e (t) represents the voltage tracking error, i L (t) represents the inductor current at the target time, i L_e (t) represents the inductor current tracking error, ∫i L_e Indicates the integral value of the inductor current tracking error.

[0049] It should be noted that V o_e (t) = V o (t)-V ref ;i L_e (t) = i L (t)-i ref , where V ref Indicates the voltage reference value, i ref Indicates the current reference value.

[0050] After obtaining the state vector, the action vector is determined according to the goal of minimizing the voltage tracking error and minimizing the inductor current tracking error.

[0051] In some embodiments of the present application, a pre-trained learning model is used to analyze the state vector, and the specific steps for obtaining the action vector are as follows: creating an initial actor network, an initial critic network, a target actor network, and a target critic network; collecting the state vectors of the boost converter model at multiple moments to form a state space, and collecting the action vectors corresponding to the state vectors at the multiple moments to form an action space; using the state vectors at multiple moments in the state space and the action vectors at multiple moments in the action space to train the initial actor network and the initial critic network to obtain the learning model; using the learning model to analyze the state vector to obtain the action vector, the action vector is used to represent the duty cycle, and the duty cycle represents the ratio of the conduction time of the switch tube in the boost converter to the switching period.

[0052] Figure 4 shows the training process of a learning model, such as Figure 4 As shown, the initialization has weights θ μ and θ q The initial actor network actorμ(s|θ μ ) and the initial critic network Q(s,a|θ q ), using weights θ μ′ and θ q′ Initializing a target actor network and a target critic network, and training the initial actor network and the initial critic network using a state vector in a state space and an action vector in an action space to obtain the learning model.

[0053] The specific process of training the initial actor network and the initial critic network using the state vectors at multiple moments in the state space and the action vectors at multiple moments in the action space to obtain the learning model is as follows: successively obtaining the state vector of each acquisition moment from the state space according to a preset step size; analyzing the state vector of each acquisition moment using the initial actor network to obtain the action vector corresponding to the state vector of each acquisition moment, and adding noise to the action vector corresponding to the state vector of each acquisition moment to obtain a first action vector; evaluating the first action vector using the initial critic network to obtain a reward corresponding to the first action vector; respectively determining the state vector of each acquisition moment, the first action vector, the reward corresponding to the first action vector, and the state vector of the next moment as experience tuples, and placing the experience tuples into a preset experience area, the state vector of the next moment represents the state vector after executing the action corresponding to the first action vector; using the experience tuples in the preset experience area to update the model parameters of the initial actor network, the initial critic network, the target actor network, and the target critic network to obtain the learning model.

[0054] It is understood that the reward function can be determined as follows: r = -5V o_e 2 -0.1a, where a represents the action vector.

[0055] like Figure 4 As shown, the process of determining the action at the target time (the first action vector) through the initial actor network according to the state vector at the target time is as follows:

[0056] a t =μ(s t |θ μ )+N t (4)

[0057] Where a t represents the action vector at the target moment, μ(s t |θ μ ) represents the state-action mapping function, θ μ represents the weight of the initial actor network, N t Indicates noise.

[0058] After determining the action at the current moment, the action at the target moment is used to operate the switch tube to obtain the state vector s at the next moment. t+1 and the reward r of the action at the target time tAn MDP (Markov Decision Process) experience tuple is determined based on the state vector at each acquisition moment, the first action vector, the reward corresponding to the first action vector, and the state vector at the next moment, and the experience tuple is placed in a preset experience area.

[0059] Among them, the process of using the experience tuples in the preset experience area to update the model parameters of the initial critic network is as follows: extracting experience tuples from the preset experience area, and using the initial actor network to predict the experience tuples to obtain the action vector of the next moment corresponding to the first action vector; adding noise to the action vector of the next moment to obtain a second action vector; using the initial critic network to evaluate the second action vector to obtain an evaluation function value of the second action vector, and the evaluation function value is used to represent the cumulative reward value of the second action vector; determining the predicted evaluation function value of the first action vector based on the evaluation function value of the second action vector; determining the loss function value based on the predicted evaluation function value of the first action vector and the actual evaluation function value of the first action vector; updating the model parameters of the initial critic network based on the loss function value.

[0060] The initial actor network is updated through policy gradient, and the update process is as follows:

[0061]

[0062] Where, represents the gradient update, J represents the objective function, β represents the strategy β, Represents the state vector s at the target time t and the action vector a at the target moment t (first action vector) expectation, where s t ~ρ β Indicates s t Under the strategy β, it conforms to the state distribution ρ,a t β indicates that the action vector at the target moment is determined by the strategy β, Q(s,a|θ q )|s=s t Denote the state vector s as s t The initial reviewer network evaluation value (Q value) at time .

[0063] The initial critic network is in state vector s t Next select action vector a t The evaluation process is as follows:

[0064]

[0065] Where, Q(st ,a t ) represents the state vector s t Next select action vector a t The predicted evaluation function value, Q(s t+1 ,a t+1 ) represents the state vector s t+1 Next select action vector a t+1 The evaluation function value of , γ represents a constant.

[0066] The loss function of the initial critic network is as follows:

[0067]

[0068] Where y t represents the predicted evaluation function value of the first action vector, Q(s t ,a t |θ q ) represents the actual evaluation function value of the first action vector, θ q represents the weight of the initial reviewer network, Represents the expected value of the difference between the predicted evaluation function value and the actual evaluation function value at the target moment, where r t ~E indicates that the reward value at the target moment is associated with the expected value.

[0069] The initial critic network is used to evaluate the second action vector to obtain an evaluation function value of the second action vector, including:

[0070] The second action vector is evaluated using the first critic network and the second critic network in the initial critic network respectively to obtain a first evaluation value and a second evaluation value; and the smaller value of the first evaluation value and the second evaluation value is determined as the evaluation function value of the second action vector.

[0071] Then, the predicted evaluation function value of the first action vector is determined according to the evaluation function value of the second action vector.

[0072] Specifically, as shown in the following formula:

[0073]

[0074] Where, represents the i-th critic network in the initial critic network (the first critic network or the second critic network in the initial critic network), ε represents noise, represents the weight of the i-th critic network in the initial critic network.

[0075] It can be understood that the evaluation process using the target critic function is the same as the initial critic network, as shown in the following formula:

[0076]

[0077] Where, represents the i-th critic network in the target critic network (the first critic network or the second critic network in the target critic network), θ μ′ represents the weight of the target actor network, represents the weight of the i-th critic network in the target critic network, μ′(s t+1 |θ μ′ ) represents the state-action mapping function of the target actor network at the next moment.

[0078] In an optional manner, the specific steps of updating the model parameters of the target reviewer network using the experience tuples in the preset experience area are as follows: in each update step, obtaining the updated model parameters of the initial reviewer network; and updating the updated model parameters to the model parameters of the target reviewer network using a soft update method.

[0079] The specific formula is as follows:

[0080]

[0081] Where τ represents the soft update coefficient, and the value range of τ is between 0 and 1.

[0082] This application also provides another control process of a boost converter, such as Figure 5 As shown, it includes: collecting current state information, including inductor current and output voltage, outer loop PID control: comparing the collected output voltage with the voltage reference value, using the PID controller to adjust the output, and generating the inductor current reference value.

[0083] Inner-loop algorithm control: After receiving the inductor current setpoint generated by the outer-loop PID controller, the inner-loop control strategy uses the TD3 algorithm (a pre-trained learning model). Based on the current state and the inductor current reference value, it generates an action (i.e., the duty cycle adjustment value) and triggers the switch via a signal generator.

[0084] To verify the feasibility of the boost converter control method proposed in this application, simulation parameters were set as a 12V DC voltage source, 1mH inductor, 100uF capacitor, 50Ω resistor, 16W constant power load, and a 24V command voltage output. The simulation duration was set to 0.35s, and initially, the boost converter command voltage output was 24V.

[0085] like Figure 6As shown in the figure, when other parameters remain unchanged, the constant power load P jumps from 24W to 36W in 0.02 seconds and drops from 36W to 30W in 0.04 seconds. Figure 6 It can be seen that at the moment of system startup, both control strategies can stabilize the output voltage at the 24V reference voltage value, but the traditional PID control exhibits significant overshoot. The proposed control strategy has better dynamic characteristics in inductor current control and reduces overshoot in voltage control. During constant power load transitions, both control strategies can continue to stabilize the output voltage at the reference value, but the proposed control strategy has smaller fluctuations and faster settling time than the dual closed-loop PID control. Figure 7 It means that when other parameters remain unchanged, the output voltage of the DC voltage source increases from 12V to 18V in 0.02 seconds. It can be seen from the figure that both control strategies can stabilize the output voltage at the reference voltage value of 24V, but the proposed control strategy has better dynamic response speed and shorter recovery time in terms of inductor current than the dual closed-loop PI control, and reduces the overshoot in voltage control.

[0086] It is understandable that Figure 6 and Figure 7 The dotted line in the figure represents the output voltage curve (PI) under traditional control, and the red line represents the output voltage curve (TD3) under the method provided in this application.

[0087] Figure 8 is a control device for a boost converter according to an embodiment of the present application, such as Figure 8 As shown, the device includes:

[0088] The acquisition module 80 is used to acquire the output voltage value and the inductor current value of the boost converter model at the target moment;

[0089] a determination module 82, configured to determine a state vector at a target moment based on the output voltage value at the target moment and the inductor current value at the target moment, wherein the state vector at the target moment is used to represent at least a deviation between the output voltage value of the boost converter at the target moment and a voltage reference value, and a deviation between the inductor current value at the target moment and an inductor current reference value, wherein the inductor current reference value is generated by a proportional-integral-derivative (PID) control loop;

[0090] The control module 84 is used to analyze the state vector using a pre-trained learning model to obtain an action vector, wherein the action vector is used to convert into a switching signal, and the switching signal is used to control the on and off of the switch tube in the boost converter to control the output voltage value of the boost converter to be equal to the voltage reference value.

[0091] In the control device of the boost converter, a PID control method is used to generate a reference inductor current, and a predetermined learning model is used to obtain the output voltage. The PID control method and the model algorithm are combined to achieve the purpose of accurately controlling the output voltage value of the boost converter, thereby achieving the technical effect of improving the control accuracy, and further solving the technical problem of low control accuracy caused by using the PID control method to control the boost converter in the related art.

[0092] The determination module 82 includes: a determination submodule, configured to determine a state vector based on the output voltage value at the target moment and the inductor current value at the target moment, including: determining a reference value of the inductor current based on the output voltage value at the target moment and the voltage reference value; determining a voltage tracking error based on the output voltage value at the target moment and the voltage reference value, and determining an inductor current tracking error based on the inductor current value at the target moment and the reference value of the inductor current; respectively obtaining the inductor current value at the target moment and an integral value of the inductor current tracking error; and determining the state vector at the target moment based on the output voltage value at the target moment, the voltage tracking error, the inductor current value at the target moment, and the integral value of the inductor current tracking error.

[0093] The control module 84 includes: an analysis submodule, which is used to analyze the state vector using a pre-trained learning model to obtain an action vector, including: creating an initial actor network, an initial critic network, a target actor network and a target critic network; collecting the state vectors of the boost converter model at multiple moments to form a state space, and collecting the action vectors corresponding to the state vectors at the multiple moments to form an action space; using the state vectors at multiple moments in the state space and the action vectors at multiple moments in the action space to train the initial actor network and the initial critic network to obtain the learning model; using the learning model to analyze the state vector to obtain the action vector, and the action vector is used to represent the duty cycle, and the duty cycle represents the ratio of the conduction time of the switch tube in the boost converter to the switching period.

[0094] The analysis submodule includes: a training unit, which is used to train the initial actor network and the initial critic network using state vectors at multiple moments in the state space and action vectors at multiple moments in the action space to obtain the learning model, including: successively obtaining the state vector of each acquisition moment from the state space according to a preset step size; analyzing the state vector of each acquisition moment using the initial actor network to obtain the action vector corresponding to the state vector of each acquisition moment, and adding noise to the action vector corresponding to the state vector of each acquisition moment to obtain a first action vector; evaluating the first action vector using the initial critic network to obtain a reward corresponding to the first action vector; respectively determining the state vector of each acquisition moment, the first action vector, the reward corresponding to the first action vector, and the state vector at the next moment as experience tuples, and placing the experience tuples into a preset experience area, the state vector at the next moment representing the state vector after executing the action corresponding to the first action vector; using the experience tuples in the preset experience area to update the model parameters of the initial actor network, the initial critic network, the target actor network, and the target critic network to obtain the learning model.

[0095] The training unit includes: an updating subunit, which is used to update the model parameters of the initial critic network using the experience tuples in the preset experience area, including: extracting experience tuples from the preset experience area, and using the initial actor network to predict the experience tuples to obtain the action vector of the next moment corresponding to the first action vector; adding noise to the action vector of the next moment to obtain a second action vector; using the initial critic network to evaluate the second action vector to obtain an evaluation function value of the second action vector, and the evaluation function value is used to represent the cumulative reward value of the second action vector; determining the predicted evaluation function value of the first action vector based on the evaluation function value of the second action vector; determining the loss function value based on the predicted evaluation function value of the first action vector and the actual evaluation function value of the first action vector; and updating the model parameters of the initial critic network based on the loss function value.

[0096] The update subunit includes: an evaluation subunit and a soft update subunit, wherein the evaluation subunit is used to use the initial critic network to evaluate the second action vector to obtain the evaluation function value of the second action vector, including: using the first critic network and the second critic network in the initial critic network to evaluate the second action vector respectively to obtain a first evaluation value and a second evaluation value; and determining the smaller value of the first evaluation value and the second evaluation value as the evaluation function value of the second action vector.

[0097] The soft update subunit is used to update the model parameters of the target reviewer network using the experience tuples in the preset experience area, including: obtaining the updated model parameters of the initial reviewer network in each update step; and updating the updated model parameters to the model parameters of the target reviewer network using a soft update method.

[0098] It should be noted that Figure 8 The control device of the boost converter shown is used to perform Figure 2 Therefore, the relevant explanations in the above-mentioned control method of the boost converter are also applicable to the control device of the boost converter, and are not repeated here.

[0099] An embodiment of the present application further provides an electronic device, comprising: a memory and a processor, wherein the memory is used to store program instructions; and the processor is connected to the memory and is used to execute the control method of the boost converter.

[0100] An embodiment of the present application further provides a non-volatile storage medium, which includes a stored computer program, wherein a device where the non-volatile storage medium is located executes the above-mentioned control method of the boost converter by running the computer program.

[0101] An embodiment of the present application further provides a computer program product, comprising computer instructions, which, when executed by a processor, implement the steps of the control method of the boost converter in the present application.

[0102] The serial numbers of the above embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0103] In the relevant embodiments of the present application, a PID control method is used to generate a reference inductor current, and a predetermined learning model is used to obtain the output voltage. The PID control method and the model algorithm are combined to achieve the purpose of accurately controlling the output voltage value of the boost converter, thereby achieving the technical effect of improving the control accuracy, and further solving the technical problems of low control accuracy caused by the use of PID control method to control the boost converter in the related art.

[0104] In the above embodiments of the present application, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0105] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only exemplary. For example, the division of the units can be a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0106] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0107] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0108] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a server or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0109] The above is only a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A control method for a boost converter, characterized in that: include: Collect the output voltage value and the inductor current value of the boost converter model at the target time; determining a state vector at a target moment based on the output voltage value at the target moment and the inductor current value at the target moment, wherein the state vector at the target moment is used to represent at least a deviation between the output voltage value of the boost converter at the target moment and a voltage reference value, and a deviation between the inductor current value at the target moment and an inductor current reference value, wherein the inductor current reference value is generated by a proportional-integral-derivative (PID) control loop; The state vector is analyzed using a pre-trained learning model to obtain an action vector, wherein the action vector is used to convert into a switching signal, and the switching signal is used to control the switching tube in the boost converter to control the output voltage value of the boost converter to be equal to the voltage reference value.

2. The method according to claim 1, characterized in that Determining a state vector according to the output voltage value at the target time and the inductor current value at the target time includes: determining a reference value of the inductor current according to the output voltage value at the target moment and the voltage reference value; determining a voltage tracking error according to the output voltage value at the target moment and the voltage reference value, and determining an inductor current tracking error according to the inductor current value at the target moment and the reference value of the inductor current; respectively obtaining the inductor current value at the target moment and the integral value of the inductor current tracking error; The state vector at the target moment is determined according to the output voltage value at the target moment, the voltage tracking error, the inductor current value at the target moment, and the integral value of the inductor current tracking error.

3. The method according to claim 1, characterized in that The state vector is analyzed using a pre-trained learning model to obtain an action vector, including: Create an initial actor network, an initial commentator network, a target actor network, and a target commentator network; collecting state vectors of the boost converter model at multiple moments to form a state space, and collecting action vectors corresponding to the state vectors at multiple moments to form an action space; The initial actor network and the initial critic network are trained using state vectors at multiple moments in the state space and action vectors at multiple moments in the action space to obtain the learning model; The state vector is analyzed using the learning model to obtain the action vector, which is used to represent the duty cycle. The duty cycle represents the ratio of the on-time of the switch tube in the boost converter to the switching period.

4. The method according to claim 3, characterized in that The initial actor network and the initial critic network are trained using state vectors at multiple moments in the state space and action vectors at multiple moments in the action space to obtain the learning model, including: Obtaining the state vector at each acquisition moment from the state space one by one according to a preset step size; Analyzing the state vector at each acquisition moment using the initial actor network to obtain an action vector corresponding to the state vector at each acquisition moment, and adding noise to the action vector corresponding to the state vector at each acquisition moment to obtain a first action vector; Using the initial critic network to evaluate the first action vector, and obtaining a reward corresponding to the first action vector; Determining the state vector at each acquisition moment, the first action vector, the reward corresponding to the first action vector, and the state vector at the next moment as experience tuples, and placing the experience tuples into a preset experience area, wherein the state vector at the next moment represents the state vector after executing the action corresponding to the first action vector; The model parameters of the initial actor network, the initial critic network, the target actor network and the target critic network are updated using the experience tuples in the preset experience area to obtain the learning model.

5. The method according to claim 4, characterized in that Updating the model parameters of the initial reviewer network using the experience tuples in the preset experience area includes: Extracting an experience tuple from the preset experience area, and using the initial actor network to predict the experience tuple to obtain an action vector at a next moment corresponding to the first action vector; adding noise to the action vector at the next moment to obtain a second action vector; Using the initial critic network to evaluate the second action vector, obtaining an evaluation function value of the second action vector, wherein the evaluation function value is used to represent a cumulative reward value of the second action vector; determining a predicted evaluation function value of the first action vector according to the evaluation function value of the second action vector; determining a loss function value according to the predicted evaluation function value of the first action vector and the actual evaluation function value of the first action vector; The model parameters of the initial critic network are updated according to the loss function value.

6. The method according to claim 5, characterized in that Evaluating the second action vector using the initial critic network to obtain an evaluation function value of the second action vector includes: respectively evaluating the second action vector using a first critic network and a second critic network in the initial critic network to obtain a first evaluation value and a second evaluation value; A smaller value between the first evaluation value and the second evaluation value is determined as an evaluation function value of the second action vector.

7. The method according to claim 5, characterized in that Updating the model parameters of the target reviewer network using the experience tuples in the preset experience area includes: At each updating step, obtaining updated model parameters of the initial reviewer network; The updated model parameters are updated to the model parameters of the target reviewer network using a soft update method.

8. A control device for a boost converter, characterized in that: include: An acquisition module, used for acquiring an output voltage value and an inductor current value of a boost converter model at a target moment; a determination module, configured to determine a state vector at a target moment based on the output voltage value at the target moment and the inductor current value at the target moment, wherein the state vector at the target moment is used to represent at least a deviation between the output voltage value of the boost converter at the target moment and a voltage reference value, and a deviation between the inductor current value at the target moment and an inductor current reference value, wherein the inductor current reference value is generated by a proportional-integral-derivative controller (PID) control loop; A control module is used to analyze the state vector using a pre-trained learning model to obtain an action vector, wherein the action vector is used to convert into a switching signal, and the switching signal is used to control the switching tube in the boost converter to control the output voltage value of the boost converter to be equal to the voltage reference value.

9. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute the control method of the boost converter according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: processor; A memory for storing instructions executable by the processor, wherein the processor is configured to execute the instructions to implement the control method of the boost converter according to any one of claims 1 to 7.

Images