An Adaptive Voltage Control Method for a Wireless Power Transfer System for Electric Vehicles

Abstract

This invention discloses an adaptive voltage control method for a wireless power transfer system in an electric vehicle. The method includes: constructing an adaptive model predictive controller (EMC) using an RBF neural network; the EMC outputs a control quantity at time k and applies it to the wireless power transfer system; the EMC adaptively adjusts the state deviation between the wireless power transfer system and the state quantity at time k+1 of a fixed model reference system, then outputs the next control quantity and applies it to the wireless power transfer system; and the method continuously collects the system's state quantity and adjusts the control quantity to achieve online adaptive control of the wireless power transfer system. This invention achieves precise control of the dynamic wireless power transfer system in an electric vehicle, significantly improving the system's dynamic response performance and anti-interference performance. In particular, it can dynamically adjust the model in the face of various external disturbances, enhancing the stability of the control system.

Description

Technical Field

[0001] This invention relates to a charging voltage control method in the field of wireless power transmission for electric vehicles, and more particularly to an adaptive voltage control method and apparatus for a wireless power transmission system for electric vehicles. Background Technology

[0002] The core challenge for electric vehicles lies in their energy storage technology. Battery packs must simultaneously meet multiple standards, including high energy density, low cost, long lifespan, and high safety. Vehicles achieving long driving ranges often require expensive and heavy battery packs, directly impacting vehicle cost and performance. The application of wireless power transfer technology in electric vehicles not only facilitates daily use but also addresses safety concerns such as potential electric shock risks associated with traditional wired charging. Dynamic wireless power transfer systems allow electric vehicles to charge while driving, improving driving range and reducing battery capacity to save costs. This solves the problem of long charging times in wired and static wireless modes, ensuring the safety and convenience of power transfer, demonstrating significant potential and becoming a crucial development direction in the future of electric vehicle charging.

[0003] In dynamic wireless power transfer systems for electric vehicles, the coupling coefficient between the transmitting and receiving coils changes as the vehicle passes between adjacent transmitting coils, causing fluctuations in the output voltage. Therefore, control algorithms are needed to stabilize the output voltage. Various control methods exist, but these existing technologies still suffer from the following problems:

[0004] First, current control methods typically include transmitter-end control and dual-end control. Transmitter-end control requires the installation of a controller at the road-end coil, which significantly increases the cost of installation, calibration, and maintenance due to the large number of transmitter coils in dynamic charging systems. Dual-end control relies on communication between the transmitter and receiver, and is susceptible to signal interference in the strong magnetic field environment generated by the energy transmission coil, making it difficult to guarantee the stability of the control system.

[0005] Second, current control methods, such as PI control and model predictive control, only regulate the output voltage through state feedback and do not fully consider the drastic changes in the coupling coefficient during dynamic wireless power transmission, which leads to limitations in the system's anti-interference performance.

[0006] Third, in current model predictive control methods, the modeling of the controlled object is fixed. However, in practical applications, the system is affected by a variety of internal and external factors, including changes in the coupling coefficient of the transceiver coil, small electronic components not considered in the modeling, and signal sampling interference. These factors may lead to a mismatch between the modeling and the system, thereby reducing the control accuracy of the system. Summary of the Invention

[0007] To address the technical challenges faced by wireless power transfer systems for electric vehicles, this invention proposes an adaptive voltage control method and system for such systems. This method controls the receiving end of the power transfer system and uses an RBF neural network to supplement and improve the mathematical model of the control circuit, predicting disturbance terms experienced by the system. This enables precise control of the dynamic wireless power transfer system of the electric vehicle, significantly improving the system's dynamic response performance and anti-interference performance. It also enhances the system's resistance to changes in the coupling coefficient of the transmitting and receiving coils under dynamic conditions. In particular, it can dynamically adjust the model in response to various external disturbances, thus enhancing the stability of the control system.

[0008] To achieve the above objectives, the present invention proposes the following technical solution:

[0009] I. An Adaptive Voltage Control Method for a Wireless Power Transfer System in an Electric Vehicle

[0010] An adaptive model predictive controller is constructed by combining an RBF neural network. The adaptive model predictive controller outputs a control quantity at time k and applies it to the wireless power transmission system of the electric vehicle. After adaptively adjusting the state deviation based on the state quantities at time k+1 of the wireless power transmission system of the electric vehicle and the fixed model reference system, the adaptive model predictive controller outputs the next control quantity and applies it to the wireless power transmission system of the electric vehicle. By continuously collecting the state quantities of the system and adjusting the control quantities, the online adaptive control of the wireless power transmission system of the electric vehicle is completed.

[0011] The adaptive model predictive controller includes a model predictive controller, a fixed model reference system, and an RBF adaptive model optimizer. The fixed model reference system estimates the state quantity of the electric vehicle wireless power transmission system at time k+1 based on the state quantity at time k of the electric vehicle wireless power transmission system and the control quantity at time k output by the model predictive controller. The state deviation is obtained by subtracting the estimated state quantity at time k+1 from the actual state quantity at time k+1. The RBF adaptive model optimizer performs single-step training based on the training samples composed of the control quantity and the corresponding state deviation and outputs the estimated state deviation. The model predictive controller combines the fixed model reference system and the RBF adaptive model optimizer to adaptively adjust the control quantity and output the next control quantity.

[0012] The RBF adaptive model optimizer trains the RBF neural network based on the state deviation calculated from the system's sampled signals during system operation. The trained RBF neural network is then updated in the model predictive controller, thereby achieving adaptive model optimization and adaptive model predictive control.

[0013] During training, gradient descent is used to optimize the network parameters of the RBF adaptive model.

[0014] The optimization objective in the model predictive controller satisfies the following formula:

[0015]

[0016]

[0017] Where N represents the step size of the prediction interval and the control interval, U k f(U) represents the predicted control sequence within the control interval. k The expression represents the optimization objective in model predictive control, where Q, P, and F represent the weight matrices for the three performance requirements: control deviation within the prediction interval, system energy output, and control deviation at the final time step, respectively. This represents the square of the second norm of the weighted matrix Q. This represents the square of the second norm of the weighted matrix P. e represents the square of the second norm of the weighted matrix F. k+i u represents the predicted value of the system control deviation at time k+i based on the system state at time k. k+i Let e ​​represent the predicted value of the system control signal at time k+i based on the system state at time k, where i = 0, ..., N-1. k+N x represents the predicted value of the system control deviation at time k+N based on the system state at time k. k+1 and x k The system state variables at time k+1 and time k are respectively, u k y represents the system input at time k. k u represents the output at time k. min and u max Represents the minimum and maximum values ​​of the input quantity, y ref e represents the expected output quantity. k Let RBF(x) be the deviation at time k. k ,u k (A) represents the compensation part of the RBF neural network in the model, A is the state matrix of the system state-space expression, B is the input matrix, C is the output matrix, and D is the pass-through matrix. T represents the discretization calculation of the matrix of a continuous state-space system. s For system sampling time, The state matrix is ​​the discretized state matrix. Let T be the discretized input matrix, and let i be the transpose. Ld Indicates the inductor L in the receiver control circuit d The current, R L L represents the load resistance at the receiving end of the power transmission system. d C represents the inductance of the receiver control circuit. d2The capacitor represents the capacitance of the receiver control circuit; μ represents the controller output signal; V in This indicates the input voltage of the receiver control circuit, V. out This represents the system output voltage that falls across the load resistor.

[0018] II. An Adaptive Voltage Control Device for a Wireless Power Transfer System in an Electric Vehicle

[0019] A fixed model reference system is used to estimate the state variables of the electric vehicle wireless power transmission system at time k+1 based on the state variables at time k and the control variables at time k output by the model predictive controller.

[0020] The RBF adaptive model optimizer is used to train the RBF neural network based on the state deviation calculated from the system's sampled signals during system operation. The trained RBF neural network is then updated in the model predictive controller, thereby achieving adaptive model optimization and adaptive model predictive control.

[0021] The model predictive controller is used to adaptively adjust the control quantity and output the next control quantity by combining a fixed model reference system and an RBF adaptive model optimizer.

[0022] III. A computer device

[0023] The device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the adaptive voltage control method for a wireless power transmission system for electric vehicles.

[0024] IV. A computer-readable storage medium

[0025] The medium stores a computer program, which, when executed by a processor, implements the steps of the adaptive voltage control method for a wireless power transmission system for electric vehicles.

[0026] V. A computer program product

[0027] The product includes a computer program / instructions that, when executed by a processor, implement the steps of the adaptive voltage control method for a wireless power transfer system for electric vehicles.

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0029] First, the present invention adopts a receiver-end control method, which installs a controller at the receiver end of the electric vehicle, avoiding the complex communication problems caused by dual-end control. Compared with installing a controller at each transmitter end, it significantly reduces installation and maintenance costs, and the controller calibration is more convenient.

[0030] Second, the present invention employs a model predictive control method, which, while obtaining state feedback, can predict the changing trend of the coupling coefficient of the transceiver coil, thereby obtaining more system information and exhibiting excellent anti-interference performance when the coupling coefficient changes drastically.

[0031] Third, this invention introduces an RBF neural network, which predicts system uncertainty information based on the phase difference between the reference system model and the real system. This allows for dynamic updates to the system modeling in model predictive control during system operation, adaptively addressing mismatches between the model and the system. This effectively improves the system's control accuracy and anti-interference performance, providing technical support for the feasibility of dynamic wireless power transfer systems for electric vehicles. It also promotes the development and application of wireless power transfer technology and is expected to have a significant impact on the future direction of electric vehicle charging. Attached Figure Description

[0032] Figure 1 This is a circuit diagram of a dynamic wireless power transfer system for electric vehicles.

[0033] Figure 2 This is a structural diagram of the proposed adaptive model predictive control method.

[0034] Figure 3 This is a diagram showing the effect of the dynamic response of the output voltage.

[0035] Figure 4 This is a diagram illustrating the output voltage response during the dynamic wireless charging process.

[0036] Figure 5 This is a diagram showing the effect of changes in the input voltage at the receiving end.

[0037] Figure 6 This is a diagram showing the effect of RBF in predicting uncertainties. Detailed Implementation

[0038] The invention will now be described in further detail with reference to the accompanying drawings. The circuit diagram of the electric vehicle dynamic wireless power transfer system designed according to the invention is shown below. Figure 1 As shown in the diagram. The energy transmitter is located on one side of the road, and the energy receiver is located at the bottom of the electric vehicle. The transmitting and receiving coils of the two will be relatively misaligned as the vehicle moves. A framework diagram of the method proposed in this invention is shown below. Figure 2 As shown, this control method utilizes an RBF neural network to refine the mathematical model of the control circuit, thereby enhancing control performance. The goal of this method is to ensure that the output voltage of the electric vehicle stably tracks the desired value during dynamic driving, and to maintain excellent performance even when the coupling coefficient of the transceiver coil changes significantly.

[0039] The input-output relationship of the receiver control circuit of the dynamic wireless power transfer system for electric vehicles can be described by the following set of differential equations:

[0040]

[0041] Among them, R L L represents the load resistance at the receiving end of the power transmission system. d C represents the inductance of the receiver control circuit. d2 V represents the capacitor of the receiver control circuit; μ represents the controller output signal; V in This represents the input voltage of the receiver control circuit, and its value can change due to external interference. (V) out This represents the system output voltage across the load resistor, and also represents the capacitance C. d2 voltage, i Ld Inductance L d The current.

[0042] The specific process of the control method is as follows:

[0043] An adaptive model predictive controller is constructed by combining an RBF neural network. The adaptive model predictive controller outputs a control quantity at time k and applies it to the wireless power transmission system of the electric vehicle. After adaptively adjusting the state deviation based on the state quantities at time k+1 of the wireless power transmission system of the electric vehicle and the fixed model reference system, the adaptive model predictive controller outputs the next control quantity and applies it to the wireless power transmission system of the electric vehicle. By continuously collecting the state quantities of the system and adjusting the control quantities, the online adaptive control of the wireless power transmission system of the electric vehicle is completed.

[0044] The adaptive model predictive controller includes a model predictive controller, a fixed model reference system, and an RBF adaptive model optimizer. The fixed model reference system estimates the state quantity of the electric vehicle wireless power transmission system at time k+1 based on the state quantity at time k of the electric vehicle wireless power transmission system and the control quantity at time k output by the model predictive controller. The state deviation is obtained by subtracting the estimated state quantity at time k+1 from the actual state quantity at time k+1. The RBF adaptive model optimizer performs single-step training based on the training samples composed of the control quantity and the corresponding state deviation and outputs the estimated state deviation. The model predictive controller combines the fixed model reference system and the RBF adaptive model optimizer to adaptively adjust the control quantity and output the next control quantity.

[0045] The RBF adaptive model optimizer trains an RBF neural network based on the state deviation calculated from the system's sampled signals during system operation. The trained RBF neural network is then updated in the model predictive controller, improving the control circuit modeling of the wireless power transmission system. Single-step training ensures that each execution of the model predictive controller uses the appropriate model for the current moment, thus achieving adaptive model optimization and adaptive model predictive control. During training, gradient descent is used to optimize the network parameters of the RBF adaptive model.

[0046] The optimization objective in the model predictive controller satisfies the following formula:

[0047]

[0048] Where N represents the step size of the prediction interval and the control interval, U k f(U) represents the predicted control sequence within the control interval. k The matrix ) represents the optimization objective in model predictive control. Q, P, and F are constant matrices, representing the weight matrices of the three performance requirements: control deviation within the prediction interval, system energy output, and control deviation at the final time step, respectively. This represents the square of the second norm of the weighted matrix Q. This represents the square of the second norm of the weighted matrix P. e represents the square of the second norm of the weighted matrix F. k+i u represents the predicted value of the system control deviation at time k+i based on the system state at time k. k+i Let e ​​represent the predicted value of the system control signal at time k+i based on the system state at time k, where i = 0, ..., N-1. k+N x represents the predicted value of the system control deviation at time k+N based on the system state at time k. k+1 and x k The system state variables at time k+1 and time k are respectively, u k y represents the system input at time k, i.e., the control input. k u represents the output at time k. min and u max Represents the minimum and maximum values ​​of the input quantity, y ref e represents the expected output quantity. k Let RBF(x) be the deviation at time k. k ,u k (A) represents the compensation part of the RBF network in the model. (B) represents the state matrix of the system state-space expression, indicating the influence of the state variables from the previous time step on the state variables from the next time step. (C) represents the input matrix, indicating the influence of the input variables on the changes in the state variables. (D) represents the output matrix, indicating the influence of the system state variables on the output variables. (F) represents the pass-through matrix, indicating the influence of the system input variables on the output variables. T represents the discretization calculation of the matrix of a continuous state-space system. s For system sampling time, The state matrix is ​​the discretized state matrix. Let T be the discretized input matrix, and let i be the transpose. Ld Indicates the inductor L in the receiver control circuit d The current, R L L represents the load resistance at the receiving end of the power transmission system.d C represents the inductance of the receiver control circuit. d2 The capacitor represents the capacitance of the receiver control circuit; μ represents the controller output signal; V in This represents the input voltage of the receiver control circuit, whose value is subject to change due to external interference; that is, it is a variable that changes with time. (V) out This represents the system output voltage across the load resistor, and also represents the capacitance C. d2 The voltage.

[0049] Simulation results are displayed on Figures 3 to 6 middle. Figure 3 This invention demonstrates the performance of a dynamic wireless power transfer system for electric vehicles in tracking rising step changes in output voltage. When the desired output voltage undergoes a rising step change, the adaptive model predictive control method proposed in this invention can respond rapidly to this change, and its overshoot and convergence time are significantly better than those of traditional PID control methods. Figure 4 This paper demonstrates the tracking performance of the system output voltage during dynamic wireless charging of electric vehicles, under conditions of significant changes in the coupling coefficient. Compared to conventional model predictive control methods, the proposed method uses an RBF neural network to compensate for model uncertainties, enabling more effective tracking of the desired output. Figure 5 The simulation presents the changes in the input voltage at the receiving end, with the system fluctuations caused by the misalignment of the transceiver coils occurring between 0.2 and 0.25 seconds. Finally, Figure 6 This paper demonstrates the accurate prediction of uncertainties in the modeling of an RBF network in a dynamic power transfer system for electric vehicles. Figures A and B represent the prediction of uncertainties in two state variables, respectively. These simulation results strongly confirm the effectiveness and superior performance of this invention and further demonstrate the application potential of the proposed adaptive model predictive control method in dynamic wireless power transfer systems for electric vehicles.

[0050] This invention proposes an adaptive voltage control device for a wireless power transfer system for electric vehicles, comprising:

[0051] A fixed model reference system unit is used to estimate the state variables of the electric vehicle wireless power transmission system at time k+1 based on the state variables at time k of the electric vehicle wireless power transmission system and the control variables at time k output by the model predictive controller.

[0052] The RBF adaptive model optimizer is used to train the RBF neural network based on the state deviation calculated from the system's sampled signals during system operation. The trained RBF neural network is then updated in the model predictive controller to improve the modeling of the wireless power transmission system control circuit. Single-step training ensures that the model predictive controller is used with the current time-to-time adaptation model each time it is executed, thereby achieving adaptive model optimization and adaptive model predictive control.

[0053] The model predictive controller is used to adaptively adjust the control quantity and output the next control quantity by combining a fixed model reference system unit and an RBF adaptive model optimizer.

[0054] This invention proposes a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of an adaptive voltage control method for a wireless power transmission system for electric vehicles.

[0055] This invention proposes a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of an adaptive voltage control method for a wireless power transmission system for electric vehicles.

[0056] This invention proposes a computer program product, including a computer program / instructions, which, when executed by a processor, implements the steps of an adaptive voltage control method for a wireless power transfer system for electric vehicles.

[0057] This invention is not limited to this embodiment. Any equivalent concept or modification within the technical scope disclosed in this invention shall be included within the protection scope of this invention.

Claims

1. An adaptive voltage control method for a wireless power transfer system for electric vehicles, characterized in that, Includes the following steps: An adaptive model predictive controller is constructed by combining an RBF neural network. The adaptive model predictive controller outputs a control quantity at time k and applies it to the wireless power transmission system of the electric vehicle. The adaptive model predictive controller adaptively adjusts the state deviation based on the state quantity at time k+1 of the wireless power transmission system of the electric vehicle and the fixed model reference system, and then outputs the next control quantity and applies it to the wireless power transmission system of the electric vehicle. The system's state quantity is continuously collected and the control quantity is adjusted, thereby completing the online adaptive control of the wireless power transmission system of the electric vehicle. The adaptive model predictive controller includes a model predictive controller, a fixed model reference system, and an RBF adaptive model optimizer. The fixed model reference system estimates the state quantity of the electric vehicle wireless power transmission system at time k+1 based on the state quantity at time k of the electric vehicle wireless power transmission system and the control quantity at time k output by the model predictive controller. The state deviation is obtained by subtracting the estimated state quantity at time k+1 from the actual state quantity at time k+1. The RBF adaptive model optimizer performs single-step training based on the training samples composed of the control quantity and the corresponding state deviation and outputs the estimated state deviation. The model predictive controller combines the fixed model reference system and the RBF adaptive model optimizer to adaptively adjust the control quantity and output the next control quantity. The RBF adaptive model optimizer trains the RBF neural network based on the state deviation calculated from the system sampled signal during system operation. The trained RBF neural network is then updated in the model predictive controller, thereby realizing adaptive model optimization and adaptive model predictive control. The optimization objective in the model predictive controller satisfies the following formula: in, Indicates the step size of the prediction interval and the control interval. This represents the predicted control sequence within the control interval. This represents the optimization objective in model predictive control. These represent the weight matrices for the three performance requirements: control deviation within the prediction interval, system energy output, and control deviation at the final time step. Represents a weighted matrix The square of the second norm, Represents a weighted matrix The square of the second norm, Represents a weighted matrix The square of the second norm, Indicates in At any given time, the system state is as follows The predicted value of the system control deviation at any given time. express At any given time, the system state is as follows The predicted value of the system control signal at any given time. , Indicates in At any given time, the system state is as follows The predicted value of the system control deviation at any given time. and They are respectively Time and System state variables at time 1. express System input at any given time express Output at any given time and This represents the minimum and maximum values ​​of the input quantity. Indicates the expected output quantity. for The deviation at time. This refers to the compensation part of the RBF neural network in the model. The state matrix is ​​the state-space representation of the system. For the input matrix, For the output matrix, For a through matrix, This represents the matrix discretization calculation of a continuous state-space system. For system sampling time, The state matrix is ​​the discretized state matrix. The input matrix is ​​discretized, and T denotes the transpose. Indicates the inductance in the receiver control circuit The current, This indicates the load resistance at the receiving end of the power transmission system. Indicates the inductance of the receiver control circuit. This refers to the capacitor in the receiver control circuit. Indicates the controller output signal; This indicates the input voltage of the receiver control circuit. This represents the system output voltage that falls across the load resistor.

2. The adaptive voltage control method for a wireless power transfer system for electric vehicles according to claim 1, characterized in that, During training, gradient descent is used to optimize the network parameters of the RBF adaptive model.

3. An adaptive voltage control device for a wireless power transfer system for electric vehicles, characterized in that, The adaptive voltage control device employs the adaptive voltage control method for a wireless power transfer system for electric vehicles as described in claim 1, and the adaptive voltage control device comprises: A fixed model reference system is used to estimate the state variables of the electric vehicle wireless power transmission system at time k+1 based on the state variables at time k and the control variables at time k output by the model predictive controller. The RBF adaptive model optimizer is used to train the RBF neural network based on the state deviation calculated from the system sampled signal during system operation. The trained RBF neural network is then updated in the model predictive controller, thereby realizing adaptive model optimization and adaptive model predictive control. The model predictive controller is used to adaptively adjust the control quantity and output the next control quantity by combining a fixed model reference system and an RBF adaptive model optimizer.

4. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the adaptive voltage control method for a wireless power transmission system for electric vehicles as described in any one of claims 1 to 2.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the adaptive voltage control method for a wireless power transmission system for electric vehicles as described in any one of claims 1 to 2.

6. A computer program product comprising a computer program / instructions, characterized in that, When executed by a processor, the computer program / instructions implement the steps of the adaptive voltage control method for a wireless power transfer system for an electric vehicle as described in any one of claims 1 to 2.

Images