Composite control method for bidirectional DC / DC converter of automobile charging pile

Abstract

The invention relates to the technical field of DC-DC converter control, in particular to a composite control method for a bidirectional DC / DC converter of an automobile charging pile. The method comprises the following steps: firstly, establishing a reduced order transfer function of the bidirectional DC / DC converter of the automobile charging pile according to a frequency sweep method, and converting the reduced order transfer function into a second order system to obtain a control model; secondly, according to the control model, an error function is established, a sliding mode surface is obtained, and a sliding mode controller is designed; meanwhile, designing an extended state observer according to the control model to obtain a disturbance estimation value; combining the sliding mode controller and the disturbance estimation value to construct a sliding mode anti-disturbance controller; then, according to the error function, establishing a neural network PID controller; and finally, according to the sliding mode anti-interference controller and the PID controller, establishing a composite controller. The composite controller can enhance the control performance of the system under complex working conditions, and the stability and safety of charging and discharging of the electric vehicle are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of DC-DC converter control, and in particular to a composite control method for a bidirectional DC / DC converter of an automobile charging pile. Background Art

[0002] As the core component of the electric vehicle charging system, the bidirectional DC / DC converter of the car charging pile is mainly used to realize the bidirectional energy transmission between the charging pile and the electric vehicle battery, and meet different voltage and current requirements in the charging and discharging modes. With the popularization of electric vehicles and the widespread deployment of charging infrastructure, the control technology of the bidirectional DC / DC converter of the car charging pile plays a vital role in realizing efficient, safe and stable energy transmission.

[0003] The bidirectional DC / DC converter of the car charging pile needs to cope with multiple complex working conditions. First, when the electric car is charging, the bidirectional DC / DC converter is responsible for stepping down the high-voltage DC power provided by the power grid to the adaptive voltage of the electric car battery to ensure the efficiency and safety of battery charging. Second, when the electric car reverse discharges to provide backup power for the power grid or home energy storage system, the converter needs to boost the low-voltage DC power of the battery to a voltage level that meets the output requirements. This bidirectional energy flow function not only enhances the flexibility and practicality of the charging pile, but also provides important support for vehicle-grid interaction and distributed energy management.

[0004] In terms of control requirements, the bidirectional DC / DC converter of the car charging pile needs to have strong adaptability to cope with various uncertainties such as grid fluctuations, changes in battery characteristics, and sudden load changes. Especially in the face of fast charging and high power output scenarios, traditional linear control methods are difficult to guarantee the response speed and control accuracy of the system when the current and voltage change greatly. Therefore, in recent years, nonlinear control methods have been widely studied in bidirectional DC / DC converters, aiming to improve the dynamic response performance and robustness of the system through a more precise adjustment mechanism.

[0005] Among them, sliding mode control and neural network PID control show good anti-interference ability and adaptability in complex environments. Sliding mode control can achieve stable control effect under large disturbances at the input and output ends due to its insensitivity to changes in system parameters. The neural network PID controller improves the accuracy and robustness of control by adaptively adjusting PID parameters, and has strong adaptability. Aiming at the control requirements of bidirectional DC / DC converters of automobile charging piles, a composite control method is proposed, which combines the advantages of sliding mode control and neural network PID control to further enhance the control performance of the system under complex working conditions and ensure the stability and safety of charging and discharging of electric vehicles. Summary of the invention

[0006] The present invention provides a composite control method for a bidirectional DC / DC converter of a car charging pile, which is used to solve the above technical problems. The method comprises the following steps: first, according to the frequency sweep method, a reduced-order transfer function of the bidirectional DC / DC converter of the car charging pile is established, the reduced-order transfer function is converted into a second-order system, and a control model is obtained according to the second-order system; secondly, according to the control model, an error function is established to obtain a sliding surface, and a sliding mode controller is established according to the error function and the sliding surface; then, according to the error function, a neural network PID controller is established; finally, a composite controller is established according to the sliding mode controller and the PID controller.

[0007] A composite control method for a bidirectional DC / DC converter of a vehicle charging pile comprises the following steps:

[0008] S1. According to the frequency sweep method, a reduced-order transfer function of the bidirectional DC / DC converter of the automobile charging pile is established, the reduced-order transfer function is converted into a second-order system, and a control model containing a total disturbance is obtained according to the second-order system;

[0009] S2. According to the control model, an error function is established to obtain a sliding surface; a sliding mode controller is designed using the error function and the sliding surface; at the same time, an extended state observer is designed according to the control model to obtain a disturbance estimate; a sliding mode anti-disturbance controller is constructed by combining the sliding mode controller and the disturbance estimate;

[0010] S3, establishing a neural network PID controller according to the error function;

[0011] S4. Establish a composite controller based on the sliding mode anti-disturbance controller and the PID controller.

[0012] Furthermore, according to the frequency sweep method, a reduced-order transfer function of the bidirectional DC / DC converter of the automobile charging pile is established, and the reduced-order transfer function is:

[0013]

[0014] Wherein, s is the Laplace operator, c0, c1, a0 and a1 are system parameters of the bidirectional DC / DC converter of the automobile charging pile, and the system parameters are obtained by MATLAB frequency sweep.

[0015] Further, the reduced-order transfer function is converted into a second-order system, and the second-order system is:

[0016]

[0017] Among them, U0 is the output voltage of the converter, u is the driving signal input to the primary switch tube, is the first-order derivative of U0, is the second-order derivative of U0, is the first-order derivative of u;

[0018] The control model obtained according to the second-order system is:

[0019]

[0020] Where b0 is the input gain and f is the total disturbance, expressed as

[0021] Furthermore, according to the control model, an error function is established to obtain a sliding surface, and the error function is:

[0022]

[0023] Wherein, e is the error function, is the desired converter output voltage;

[0024] The sliding surface is:

[0025]

[0026] Where s is the sliding surface of the Boost converter, α is the sliding surface parameter, is the first-order derivative of e.

[0027] Furthermore, a sliding mode controller is established according to the error function and the sliding mode surface, and the sliding mode controller is:

[0028]

[0029] Among them, u s1 is the value of the sliding mode controller, β1 and β2 are controller parameters, sgn(*) is the sign function, for The second derivative of .

[0030] Furthermore, according to the control model, an extended state observer is designed, and the extended state observer is:

[0031]

[0032] Wherein, z1, z2 and z3 are the state variables of the extended state observer, and z3 is the disturbance estimate, are the first-order derivatives of z1, z2, and z3 respectively, and K1, K2, and K3 are the parameters of the extended state observer.

[0033] Further, according to the sliding mode controller and the disturbance estimation value, a sliding mode anti-disturbance controller is obtained, and the sliding mode anti-disturbance controller is:

[0034]

[0035] Among them, u s2 is the value of the sliding mode disturbance rejection controller.

[0036] Furthermore, according to the error function, a neural network PID controller is established, and the neural network PID controller is:

[0037]

[0038] Among them, γ p (k), γ p (k) and γ p (k) is the PID controller parameter at time k, and the formula is:

[0039] γ p (k) = γ p (k-1)+Δγ p (k);

[0040] γ i (k) = γ i (k-1)+Δγ i (k);

[0041] γ d (k) = γ d (k-1)+Δγ d (k);

[0042] Among them, γ p (k-1),γ i (k-1) and γ d (k-1) is the PID controller parameter at time k-1, Δγ p (k), Δγ i (k) and Δγ d (k) is the parameter increment value of the neural network PID controller at time k.

[0043] Furthermore, the neural network PID controller parameter increment value is calculated by the neural network, and the formula is:

[0044]

[0045] w k =w k-1 -η k-1 g k-1 ;

[0046]

[0047]

[0048] Among them, w k is the learning rate of the neural network at time k, E is the evaluation index, y(k) is the output of the neural network at time k, Δu(k) is the input of the neural network at time k, η k-1 is the adaptive learning rate at k-1 time, g k-1 Update the gradient for k-1 time, g i Update the gradient at time i, η 0 is the initial adaptive learning rate, σ k-1 is the cumulative gradient square sum at time k-1, e(k) is the tracking error at time k, e(k-1) is the tracking error at time k-1, and e(k-2) is the tracking error at time k-2. is the expected converter output voltage at time k, and U0(k) is the actual converter output voltage at time k.

[0049] Furthermore, a composite controller is established according to the sliding mode controller and the PID controller, and the composite controller is:

[0050] u=λu PID +(1-λ)u s2 ;

[0051] Among them, λ is the weight, and the formula is:

[0052]

[0053] Among them, e a is the first threshold, e b is the second threshold.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] 1. The frequency response of the system can be quickly obtained by using the frequency sweep method, thereby establishing a reduced-order transfer function and simplifying the high-order complexity of the original system. After the reduced-order transfer function is converted into a second-order system, the complexity of the control model can be reduced while maintaining the dynamic characteristics of the system, making the controller design and parameter debugging more efficient and reducing the time cost of system development.

[0056] 2. The sliding mode controller uses the sliding surface to constrain the system state and has strong anti-interference characteristics; while the extended state observer can estimate the unknown disturbances and uncertainties in the system in real time, and feed back the estimated values ​​to the controller, so that the controller can automatically compensate for the deviations caused by disturbances. By combining the sliding mode controller with the extended state observer, a sliding mode anti-disturbance controller is designed, which can not only reduce the fluctuations of the system under disturbances, but also improve the control accuracy and robustness of the system.

[0057] 3. A neural network PID controller was designed, and an adaptive learning rate was used to update the parameter increment value of the neural network PID controller, so that the controller can dynamically adjust the learning rate in real time to ensure rapid convergence when the error is large and smooth adjustment when the error is small. This mechanism significantly improves the convergence speed and accuracy of the controller and avoids the system oscillation problem that may be caused by the traditional fixed learning rate.

[0058] 4. The composite controller in the present invention integrates the fast response and robustness of sliding mode control and the adaptive adjustment capability of neural network PID, which enables the system to achieve optimized control under complex working conditions, ensures the stability and high precision of the system, and enables the system to dynamically balance the charging and discharging requirements and energy management, thereby significantly improving the reliability and control effect of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a schematic diagram of the process of the present invention;

[0060] Figure 2 It is the circuit diagram of the bidirectional DC / DC converter for the car charging pile;

[0061] Figure 3 It is the block diagram of the composite control system of the bidirectional DC / DC converter of the automobile charging pile. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] In modern energy management, the control accuracy and reliability of the bidirectional DC / DC converter of a car charging pile are crucial to the overall performance of the system. Although traditional control technology performs well in some scenarios, it is often difficult to fully meet the system requirements under complex energy environments and dynamic conditions. To this end, the present invention provides a composite control method for a bidirectional DC / DC converter of a car charging pile. The method combines sliding mode control and neural network PID control, and through the real-time control mechanism of the bidirectional DC / DC converter, utilizes the fast response and anti-interference performance of the sliding mode control, and the adaptive adjustment ability of the neural network PID, to achieve efficient optimization of energy flow and power scheduling. Through this composite control strategy, the bidirectional DC / DC converter of the car charging pile can maintain excellent control performance in a changing operating environment, ensure the stability and reliability of the system under different working conditions, and improve the operating efficiency of the entire energy management system.

[0064] Embodiment 1

[0065] In the embodiment of the present application, a composite control method of a bidirectional DC / DC converter of a vehicle charging pile is adopted; the specific implementation process is as follows Figure 1 As shown, first, according to the sweep frequency method, a reduced-order transfer function of the bidirectional DC / DC converter of the automobile charging pile is established, and the reduced-order transfer function is converted into a second-order system, and a control model containing a total disturbance is obtained according to the second-order system; secondly, according to the control model, an error function is established to obtain a sliding surface; using the error function and the sliding surface, a sliding mode controller is designed; at the same time, an extended state observer is designed according to the control model to obtain a disturbance estimate; combining the sliding mode controller and the disturbance estimate, a sliding mode anti-disturbance controller is constructed; then, according to the error function, a neural network PID controller is established; finally, a composite controller is established according to the sliding mode anti-disturbance controller and the PID controller. This embodiment discusses the control strategy of the bidirectional DC / DC converter of the automobile charging pile using a composite control strategy of BP neural network PID control combined with sliding mode anti-disturbance control under the condition of constant expected voltage input.

[0066] Furthermore, the specific implementation process of the embodiment of the present application is as follows:

[0067] The circuit diagram of the bidirectional DC / DC converter of the car charging pile is as follows Figure 2 As shown, the circuit uses an H-bridge structure at each end, and the input and output isolation is achieved through a high-frequency transformer. There is an input voltage U at the input and output ends respectively. in And output voltage U0, with input capacitor C in The input H-bridge is composed of four switching devices Q1, Q2, Q3 and Q4, diodes D1, D2, D3 and D4, and capacitors C1, C2, C3 and C4. The output H-bridge is composed of four switching devices Q5, Q6, Q7 and Q8, diodes D5, D6, D7 and D8, and capacitors C5, C6, C7 and C8. The function of the H-bridge structure is to convert DC power into a high-frequency AC signal for transmission through a transformer. Capacitors C1, C2, C3, C4, C5, C6, C7 and C8 are used for filtering and voltage stabilization. Resonant inductor L r1 , L r2 and leakage inductance L m Forming a soft switching circuit, and resonant capacitor C r1 , C r2Together they form a resonant circuit, thereby achieving zero voltage or zero current switching, reducing switching losses and improving system efficiency. Transformer T achieves electrical isolation between input and output, and converts the voltage up and down according to the turns ratio of the winding. By controlling the on and off of the input and output H-bridge switches, the circuit can achieve bidirectional energy flow at both ends, that is, energy can be transferred from the input to the output when needed, and from the output back to the input when reversed.

[0068] The block diagram of the composite control system of the bidirectional DC / DC converter of the automobile charging pile is as follows: Figure 3 As shown in the figure, it consists of a neural network, a PID controller, a sliding mode anti-disturbance controller and an extended state observer. Each component cooperates with each other to achieve precise control of the output voltage U0 so that it can stably follow the desired voltage. The neural network generates PID parameter increments Δγ p , Δγ i and Δγ d , the parameters of the PID controller are adaptively adjusted, and the PID controller generates a control signal u according to the adjusted parameters PID The sliding mode anti-disturbance controller is based on the sliding mode control theory and is used to suppress the uncertainty and external disturbance in the system. The generated control signal u s2 The extended state observer is responsible for real-time estimation of the system state and disturbance information, providing support for the sliding mode anti-disturbance controller to compensate for disturbances more accurately. The final control signal u is the output u of the PID controller. PID The output u of the sliding mode disturbance rejection controller s2 Synthesized and applied to the bidirectional DC / DC converter of the car charging pile.

[0069] Furthermore, according to the frequency sweep method, a reduced-order transfer function of the bidirectional DC / DC converter of the automobile charging pile is established, and the reduced-order transfer function is:

[0070]

[0071] Where s is the Laplace operator, c0, c1, a0 and a1 are the system parameters of the bidirectional DC / DC converter of the automobile charging pile, and the system parameters are obtained by MATLAB frequency sweep, which is c0=-2.73×10 5 , c1=0.87, a0=8.48×10 8 , a1=783.5.

[0072] In the embodiment of the present application, the sweep frequency method is used to obtain the system transfer function. The sweep frequency method can quickly obtain the frequency response of the system, thereby establishing a reduced-order transfer function and simplifying the high-order complexity of the original system. After the reduced-order transfer function is converted into a second-order system, the complexity of the control model can be reduced while maintaining the dynamic characteristics of the system, making the controller design and parameter debugging more efficient and reducing the time cost of system development.

[0073] Further, the reduced-order transfer function is converted into a second-order system, and the second-order system is:

[0074]

[0075] Among them, U0 is the output voltage of the converter, u is the driving signal input to the primary switch tube, is the first-order derivative of U0, is the second-order derivative of U0, is the first-order derivative of u;

[0076] The control model obtained according to the second-order system is:

[0077]

[0078] Where b0 is the input gain and f is the total disturbance, expressed as

[0079] In the embodiment of the present application, by converting the frequency domain model into a time domain control model, the dynamic response characteristics of the system can be analyzed more intuitively, and the response speed and stability of the system to disturbances can be improved, thereby achieving precise control of the converter output voltage. This time domain modeling method effectively improves the robustness, adaptability and dynamic performance of the system, ensuring reliability and efficiency under various working conditions.

[0080] Furthermore, according to the control model, an error function is established to obtain a sliding surface, and the error function is:

[0081]

[0082] Wherein, e is the error function, is the desired converter output voltage, define Then the error function is:

[0083] e=40-U0;

[0084] The sliding surface is:

[0085]

[0086] Where s is the sliding surface of the Boost converter, α is the sliding surface parameter, is the first-order derivative of e.

[0087] In the embodiment of the present application, by establishing an error function and defining a sliding surface according to the control model, the system can accurately describe the deviation between the converter output voltage and the desired value, thereby achieving effective feedback and control of the system deviation. This design enables the controller to quickly generate an adjustment signal and guide the state to converge to the desired value along the sliding surface when the output voltage deviates from the desired value, thereby achieving precise control of the converter output voltage.

[0088] Furthermore, a sliding mode controller is established according to the error function and the sliding mode surface, and the sliding mode controller is:

[0089]

[0090] Among them, u s1 is the value of the sliding mode controller, β1 and β2 are the controller parameters, and sgn(*) is the sign function.

[0091] In the embodiment of the present application, by introducing the sign function and the second-order derivative of the error, the sliding mode controller can quickly respond to the deviation of the system state and guide the system back to the sliding mode surface with precise control quantity, thereby effectively improving the dynamic tracking capability and response speed of the system and realizing rapid regulation of the output voltage.

[0092] Furthermore, according to the control model, an extended state observer is designed, and the extended state observer is:

[0093]

[0094] Wherein, z1, z2 and z3 are the state variables of the extended state observer, and z3 is the disturbance estimate, are the first-order derivatives of z1, z2, and z3 respectively, and K1, K2, and K3 are the parameters of the extended state observer.

[0095] In the embodiment of the present application, by designing an extended state observer, the unknown disturbance in the system can be effectively estimated in real time. The extended state observer uses the dynamic adjustment of state variables to track the disturbance, so that the system can obtain and compensate for the error caused by the disturbance in a timely manner, thereby enhancing the real-time and accuracy of the control.

[0096] Further, according to the sliding mode controller and the disturbance estimation value, a sliding mode anti-disturbance controller is obtained, and the sliding mode anti-disturbance controller is:

[0097]

[0098] Among them, u s2 is the value of the sliding mode disturbance rejection controller.

[0099] In the embodiment of the present application, by combining the sliding mode controller with the extended state observer, a sliding mode anti-disturbance controller is designed, which can realize real-time compensation and suppression of external disturbances, significantly reduce the output fluctuation of the system under various disturbance conditions, and improve the stability of the system. The sliding mode anti-disturbance controller uses the disturbance estimation value obtained by the extended state observer for dynamic compensation, so that the system can respond quickly and adapt to changes in the external environment.

[0100] Furthermore, according to the error function, a neural network PID controller is established, and the neural network PID controller is:

[0101]

[0102] Among them, γ p (k), γ p (k) and γ p (k) is the PID controller parameter at time k, and the formula is:

[0103] γ p (k) = γ p (k-1)+Δγ p (k);

[0104] γ i (k) = γ i (k-1)+Δγ i (k);

[0105] γ d (k) = γ d (k-1)+Δγ d (k);

[0106] Among them, γ p (k-1),γ i (k-1) and γ d (k-1) is the PID controller parameter at time k-1, βγ p (k), βγ i (k) and Δγ d (k) is the parameter increment value of the neural network PID controller at time k.

[0107] In the embodiment of the present application, the neural network PID controller combines the self-learning ability of the neural network and the fast response characteristics of the PID control. By adjusting the incremental value of the PID parameters, it can dynamically adapt to changes in the system state and optimize the control effect in real time. The controller can automatically adjust the PID parameters under various working conditions, so that the system can maintain high-precision control under different working conditions, thereby improving the response speed and control accuracy of the system.

[0108] Furthermore, the parameter increment value of the neural network PID controller is calculated by a neural network. The neural network adopts a feedforward neural network structure, which is specifically configured as a three-layer structure, including an input layer, a hidden layer and an output layer; the input layer is the tracking error e, and the input information is used to describe the current dynamic state of the system and provide a basis for the neural network to adjust the parameters; the hidden layer is configured with three neurons for processing the nonlinear mapping of the input data, and the activation function adopts the sigmoid function; the output layer directly gives the output of the PID controller parameter increment, including Δγ p (k), Δγ i (k) and Δγ d (k); The weight and learning rate formula of the neural network is:

[0109]

[0110] w k =w k-1 -η k-1 g k-1 ;

[0111]

[0112] Among them, w k is the weight of the neural network at time k, E is the evaluation index, y(k) is the output of the neural network at time k, Δu(k) is the input of the neural network at time k, η k-1 is the adaptive learning rate at k-1 time, g k-1 Update the gradient for k-1 time, g i Update the gradient at time i, η 0 is the initial adaptive learning rate, σ k-1 is the cumulative gradient square sum at time k-1, e(k) is the tracking error at time k, e(k-1) is the tracking error at time k-1, and e(k-2) is the tracking error at time k-2. is the expected converter output voltage at time k, and U0(k) is the actual converter output voltage at time k.

[0113] In the embodiment of the present application, the parameter increment value of the neural network PID controller is updated by using an adaptive learning rate, so that the controller can dynamically adjust the learning rate in real time. The adaptive learning rate automatically increases or decreases according to the size of the error and gradient, ensuring rapid convergence when the error is large and smooth adjustment when the error is small. This mechanism significantly improves the convergence speed and accuracy of the controller and avoids the system oscillation problem that may be caused by the traditional fixed learning rate.

[0114] Furthermore, a composite controller is established according to the sliding mode controller and the PID controller, and the composite controller is:

[0115] u=λuPID +(1-λ)u s2 ;

[0116] Among them, λ is the weight, and the formula is:

[0117]

[0118] Among them, e a is the first threshold, indicating the lower limit of the set error e, e b is the second threshold, indicating the upper limit of the set error e. When λ=1, it is PID control; when λ=0, it is sliding mode anti-interference control; when 0<λ<1, it is compound control.

[0119] In the embodiment of the present application, the composite controller integrates the fast response and robustness of sliding mode control and the adaptive adjustment capability of neural network PID, which enables the system to achieve optimized control under complex working conditions, ensures the stability and high precision of the system, and enables the system to dynamically balance charging and discharging requirements and energy management, thereby significantly improving the reliability and control effect of the system.

[0120] Embodiment 2

[0121] This embodiment discusses the control strategy of the bidirectional DC / DC converter of the automobile charging pile under the condition of step desired voltage input, using a composite control strategy of RBF neural network PID control combined with sliding mode anti-disturbance control.

[0122] Furthermore, according to the frequency sweep method, a reduced-order transfer function of the bidirectional DC / DC converter of the automobile charging pile is established, and the reduced-order transfer function is:

[0123]

[0124] Where s is the Laplace operator, c0, c1, a0 and a1 are the system parameters of the bidirectional DC / DC converter of the automobile charging pile, and the system parameters are obtained by MATLAB frequency sweep, which is c0=-2.73×10 5 , c1=0.87, a0=8.48×10 8 , a1=783.5.

[0125] Further, the reduced-order transfer function is converted into a second-order system, and the second-order system is:

[0126]

[0127] Among them, U0 is the output voltage of the converter, u is the driving signal input to the primary switch tube, is the first-order derivative of U0, is the second-order derivative of U0, is the first-order derivative of u;

[0128] The control model obtained according to the second-order system is:

[0129]

[0130] Where b0 is the input gain and f is the total disturbance, expressed as

[0131] Furthermore, according to the control model, an error function is established to obtain a sliding surface, and the error function is:

[0132]

[0133] Wherein, e is the error function, is the desired converter output voltage, define

[0134] The sliding surface is:

[0135]

[0136] Where s is the sliding surface of the Boost converter, α is the sliding surface parameter, is the first-order derivative of e.

[0137] Furthermore, a sliding mode controller is established according to the error function and the sliding mode surface, and the sliding mode controller is:

[0138]

[0139] Among them, u s1 is the value of the sliding mode controller, β1 and β2 are controller parameters, sgn(*) is the sign function, for The second derivative of .

[0140] Furthermore, according to the control model, an extended state observer is designed, and the extended state observer is:

[0141]

[0142] Wherein, z1, z2 and z3 are the state variables of the extended state observer, and z3 is the disturbance estimate, are the first-order derivatives of z1, z2, and z3 respectively, and K1, K2, and K3 are the parameters of the extended state observer.

[0143] Further, according to the sliding mode controller and the disturbance estimation value, a sliding mode anti-disturbance controller is obtained, and the sliding mode anti-disturbance controller is:

[0144]

[0145] Among them, u s2 is the value of the sliding mode disturbance rejection controller.

[0146] Furthermore, according to the error function, a neural network PID controller is established, and the neural network PID controller is:

[0147]

[0148] Among them, γ p (k), γ p (k) and γ p (k) is the PID controller parameter at time k, and the formula is:

[0149] γ p (k) = γ p (k-1)+Δγ p (k);

[0150] γ i (k) = γ i (k-1)+Δγ i (k);

[0151] γ d (k) = γ d (k-1)+Δγ d (k);

[0152] Among them, γ p (k-1),γ i (k-1) and γ d (k-1) is the PID controller parameter at time k-1, Δγ p (k), Δγ i (k) and Δγ d (k) is the parameter increment value of the neural network PID controller at time k.

[0153] Furthermore, the parameter increment value of the neural network PID controller is calculated by a neural network, and the neural network adopts a radial basis neural network structure, which is specifically configured as a three-layer structure, including an input layer, a hidden layer and an output layer; the input layer is the tracking error e, and the input information is used to describe the current dynamic state of the system and provide a basis for the neural network to adjust the parameters; the hidden layer uses a Gaussian function as an activation function, including three neurons, each neuron selects a center point through a clustering algorithm, and calculates the output according to the distance between the input data and these centers; the output layer directly gives the output of the PID controller parameter increment, including Δγ p (k), Δγ i (k) and Δγ d (k); The weight and learning rate formula of the neural network is:

[0154]

[0155]

[0156] w k =w k-1 -η k-1 g k-1 ;

[0157]

[0158] Among them, w k is the weight of the neural network at time k, E is the evaluation index, y(k) is the output of the neural network at time k, Δu(k) is the input of the neural network at time k, η k-1 is the adaptive learning rate at k-1 time, g k-1 Update the gradient for k-1 time, g i Update the gradient at time i, η 0 is the initial adaptive learning rate, σ k-1 is the cumulative gradient square sum at time k-1, e(k) is the tracking error at time k, e(k-1) is the tracking error at time k-1, and e(k-2) is the tracking error at time k-2. is the expected converter output voltage at time k, and U0(k) is the actual converter output voltage at time k.

[0159] Furthermore, a composite controller is established according to the sliding mode controller and the PID controller, and the composite controller is:

[0160] u=λu PID +(1-λ)u s2 ;

[0161] Among them, λ is the weight, and the formula is:

[0162]

[0163] Among them, e a is the first threshold, indicating the lower limit of the set error e, e b is the second threshold, indicating the upper limit of the set error e. When λ=1, it is PID control; when λ=0, it is sliding mode anti-interference control; when 0<λ<1, it is compound control.

[0164] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A composite control method for a bidirectional DC / DC converter of a vehicle charging pile, characterized in that: The following steps are involved: S1. According to the frequency sweep method, a reduced-order transfer function of the bidirectional DC / DC converter of the automobile charging pile is established, the reduced-order transfer function is converted into a second-order system, and a control model containing a total disturbance is obtained according to the second-order system; S2. According to the control model, an error function is established to obtain a sliding surface; Using the error function and the sliding surface, a sliding mode controller is designed; at the same time, an extended state observer is designed according to the control model to obtain a disturbance estimate; combining the sliding mode controller and the disturbance estimate, a sliding mode anti-disturbance controller is constructed; S3, establishing a neural network PID controller according to the error function; S4. Establish a composite controller based on the sliding mode anti-disturbance controller and the PID controller.

2. The composite control method of a bidirectional DC / DC converter for a vehicle charging pile according to claim 1, characterized in that: According to the frequency sweep method, a reduced-order transfer function of the bidirectional DC / DC converter of the automobile charging pile is established, and the reduced-order transfer function is: Wherein, s is the Laplace operator, c0, c1, a0 and a1 are system parameters of the bidirectional DC / DC converter of the automobile charging pile, and the system parameters are obtained by MATLAB frequency sweep.

3. The composite control method of a bidirectional DC / DC converter for a vehicle charging pile according to claim 2 is characterized in that: The reduced-order transfer function is converted into a second-order system, which is: Among them, U0 is the output voltage of the converter, u is the driving signal input to the primary switch tube, is the first-order derivative of U0, is the second-order derivative of U0, is the first-order derivative of u; The control model obtained according to the second-order system is: Where b0 is the input gain and f is the total disturbance, expressed as 4. The composite control method of a bidirectional DC / DC converter for a vehicle charging pile according to claim 3 is characterized in that: According to the control model, an error function is established to obtain a sliding surface. The error function is: Wherein, e is the error function, is the desired converter output voltage; The sliding surface is: Where s is the sliding surface of the Boost converter, α is the sliding surface parameter, is the first-order derivative of e.

5. The composite control method of a bidirectional DC / DC converter for a vehicle charging pile according to claim 4 is characterized in that: According to the error function and the sliding surface, a sliding mode controller is established, and the sliding mode controller is: Among them, u s1 is the value of the sliding mode controller, β1 and β2 are controller parameters, sgn(*) is the sign function, for The second derivative of .

6. The composite control method of a bidirectional DC / DC converter for a vehicle charging pile according to claim 3 is characterized in that: According to the control model, an extended state observer is designed, and the extended state observer is: Wherein, z1, z2 and z3 are the state variables of the extended state observer, and z3 is the disturbance estimate, are the first-order derivatives of z1, z2, and z3 respectively, and K1, K2, and K3 are the parameters of the extended state observer.

7. The composite control method of a bidirectional DC / DC converter for a vehicle charging pile according to claim 1, characterized in that: According to the sliding mode controller and the disturbance estimation value, a sliding mode anti-disturbance controller is obtained, and the sliding mode anti-disturbance controller is: Among them, u s2 is the value of the sliding mode disturbance rejection controller.

8. The composite control method of a bidirectional DC / DC converter for a vehicle charging pile according to claim 1, characterized in that: According to the error function, a neural network PID controller is established, and the neural network PID controller is: Among them, γ p (k), γ p (k) and γ p (k) is the PID controller parameter at time k, and the formula is: c p (k)=γ p (k-1)+Δγ p (k); c i (k)=γ i (k-1)+Δγ i (k); c d (k)=γ d (k-1)+Δγ d (k); Among them, γ p (k-1),γ i (k-1) and γ d (k-1) is the PID controller parameter at time k-1, Δγ p (k), Δγ i (k) and Δγ d (k) is the parameter increment value of the neural network PID controller at time k.

9. A composite control method for a bidirectional DC / DC converter of a vehicle charging pile according to claim 8, characterized in that: The parameter increment value of the neural network PID controller is calculated by the neural network, and the formula is: w k =w k-1 -or k-1 g k-1 ; Among them, w k is the learning rate of the neural network at time k, E is the evaluation index, y(k) is the output of the neural network at time k, Δu(k) is the input of the neural network at time k, η k-1 is the adaptive learning rate at k-1 time, g k-1 Update the gradient for k-1 time, g i Update the gradient at time i, η 0 is the initial adaptive learning rate, σ k-1 is the cumulative gradient square sum at time k-1, e(k) is the tracking error at time k, e(k-1) is the tracking error at time k-1, and e(k-2) is the tracking error at time k-2. is the expected converter output voltage at time k, and U0(k) is the actual converter output voltage at time k.

10. The composite control method of a bidirectional DC / DC converter of a vehicle charging pile according to claim 1, characterized in that: According to the sliding mode controller and the PID controller, a composite controller is established, and the composite controller is: u=λu PID +(1-λ)u s2 ; Among them, λ is the weight, and the formula is: Among them, e a is the first threshold, e b is the second threshold.

Images