Control method, device, system and storage medium of dual active full-bridge converter

Abstract

This application relates to a control method, apparatus, system, and computer-readable storage medium for a dual active full-bridge converter. The method includes: acquiring the current operating parameters of the dual active full-bridge converter in the current control cycle; determining the reference value of the outer phase shift angle of the dual active full-bridge converter based on the operating parameters and a preset control algorithm; determining the compensation value of the outer phase shift angle of the dual active full-bridge converter based on the operating parameters and a preset neural network model, wherein the neural network model is determined based on historical operating parameters and corresponding optimized compensation values ​​of the dual active full-bridge converter under different operating conditions; and controlling the operation of the dual active full-bridge converter based on the reference value and the compensation value. This method is beneficial for improving the stability of the dual active full-bridge converter under different operating conditions.

Description

Technical Field

[0001] This application relates to the field of AC power transmission technology, and in particular to a control method, apparatus, dual active full-bridge converter system, and computer-readable storage medium for a dual active full-bridge converter. Background Technology

[0002] A dual active bridge (DAB) converter is a bidirectional DC-DC converter consisting of a primary-side full-bridge, a high-frequency transformer, and a secondary-side full-bridge, capable of converting DC voltage. By controlling the phase difference (i.e., phase shift angle) between the drive signals of the switching transistors in the primary and secondary full-bridges of the DAB converter, the voltage conversion power of the DAB converter can be changed.

[0003] The main control algorithm currently used for DAB converters is the proportional-integral (PI) control algorithm. This algorithm collects the deviation between the actual output voltage of the DAB converter and the reference voltage, calculates the phase shift angle using the PI control algorithm, and adjusts the external phase shift angle between the drive signals of the primary and secondary full-bridge circuits of the DAB converter based on the calculated phase shift angle to control the DC-DC conversion power of the converter, thereby stabilizing the output voltage of the converter at the target value.

[0004] However, in actual operating conditions, the input voltage and load of the DAB converter may experience sudden changes. When the operating conditions change or the device parameters drift, the phase shift angle calculated using a fixed PI control algorithm will deviate, thus affecting the performance of the DAB converter. Summary of the Invention

[0005] Therefore, it is necessary to provide a control method, device, system, computer-readable storage medium, and computer program product for a dual active full-bridge converter that can improve the performance of the dual active full-bridge converter, in order to address the above-mentioned technical problems.

[0006] Firstly, this application provides a control method for a dual active full-bridge converter, including:

[0007] In the current control cycle, obtain the current operating parameters of the dual active full-bridge converter;

[0008] The reference value of the external phase shift angle of the dual active full-bridge converter is determined based on the operating parameters and the preset control algorithm;

[0009] The external phase shift compensation value of the dual active full-bridge converter is determined based on the operating parameters and the preset neural network model. The neural network model is determined based on the historical operating parameters and corresponding optimized compensation values ​​of the dual active full-bridge converter under different operating conditions.

[0010] The operation of the dual active full-bridge converter is controlled based on the external phase angle reference value and the external phase angle compensation value.

[0011] In one embodiment, the method further includes:

[0012] Obtain the historical operating parameters and corresponding optimized compensation values ​​of the dual active full-bridge converter under different operating conditions. The performance index parameters of the dual active full-bridge converter corresponding to the optimized compensation value meet the preset optimization conditions.

[0013] Using historical operating parameters as samples and optimized compensation values ​​as labels, we train a pre-set base model to obtain an intermediate model.

[0014] The intermediate model parameters are updated according to the preset descent gradient, and the loss value of the intermediate model is determined based on the preset loss function. When the loss value is less than or equal to the preset loss threshold, the current intermediate model is used as a neural network model.

[0015] In one embodiment, controlling the operation of the dual active full-bridge converter based on the outer phase shift reference value and the outer phase shift compensation value includes:

[0016] The outer phase angle control value is obtained by summing the outer phase angle reference value and the outer phase angle compensation value;

[0017] The operation of the dual active full-bridge converter is controlled based on the external phase shift control value.

[0018] In one embodiment, after controlling the operation of the dual active full-bridge converter based on the external phase shift reference value and the external phase shift compensation value, the method further includes:

[0019] Obtain the current stress of the inductor and the voltage conversion efficiency of the dual active full-bridge converter;

[0020] If the increase in current stress is within a preset first range, the neural network model is updated based on the current stress; and / or, if the decrease in voltage conversion efficiency is within a preset second range, the neural network model is updated based on the voltage conversion efficiency; the updated neural network model is used as the preset neural network model in the next control cycle.

[0021] In one embodiment, after controlling the operation of the dual active full-bridge converter based on the external phase shift reference value and the external phase shift compensation value, the method further includes:

[0022] Obtain the input voltage of the dual active full-bridge converter and the soft-switching state of the power devices in the dual active full-bridge converter;

[0023] When the input voltage is within a preset fault range and / or the soft switching state is in the case of soft switching failure, the dual active full-bridge converter is controlled based on the external phase angle reference value.

[0024] In one embodiment, after controlling the operation of the dual active full-bridge converter based on the external phase shift reference value and the external phase shift compensation value, the method further includes:

[0025] Obtain the device temperature of the power devices in the dual active full-bridge converter and the load operating status of the load connected to the dual active full-bridge converter;

[0026] When the load is short-circuited and / or the device temperature is greater than or equal to a preset temperature threshold, the dual active full-bridge converter is powered off.

[0027] In one embodiment, the operating parameters include output voltage and inductor current values; the control algorithm includes an outer voltage loop algorithm and an inner current loop algorithm; the outer phase shift reference value of the dual active full-bridge converter is determined based on the operating parameters and the preset control algorithm, including:

[0028] The voltage deviation value is determined based on the output voltage and the preset reference voltage;

[0029] The inductor current setpoint of the dual active full-bridge converter is determined based on the voltage deviation value and the voltage outer loop algorithm;

[0030] Determine the current deviation value based on the inductor current setpoint and the inductor current value;

[0031] The reference value of the outer phase shift angle of the dual active full-bridge converter is determined based on the current deviation value and the current inner loop algorithm.

[0032] Secondly, this application also provides a control device for a dual active full-bridge converter, comprising:

[0033] The data acquisition module is used to acquire the current operating parameters of the dual active full-bridge converter during the current control cycle.

[0034] The reference control module is used to determine the reference value of the external phase shift angle of the dual active full-bridge converter based on the operating parameters and the preset control algorithm.

[0035] The compensation module is used to determine the external phase shift compensation value of the dual active full-bridge converter based on the operating parameters and the preset neural network model. The neural network model is determined based on the historical operating parameters of the dual active full-bridge converter under different operating conditions and the corresponding optimized compensation values.

[0036] The control module is used to control the operation of the dual active full-bridge converter based on the external phase angle reference value and the external phase angle compensation value.

[0037] Thirdly, this application also provides a dual active full-bridge converter system, including a dual active full-bridge converter and a controller connected to each other;

[0038] The dual active full-bridge converter is used to input DC voltage, convert the input DC voltage to a transformed DC voltage, and outputs it. The controller stores a computer program, and when the controller executes the computer program, it performs the following steps:

[0039] In the current control cycle, obtain the current operating parameters of the dual active full-bridge converter;

[0040] The reference value of the external phase shift angle of the dual active full-bridge converter is determined based on the operating parameters and the preset control algorithm;

[0041] The external phase shift compensation value of the dual active full-bridge converter is determined based on the operating parameters and the preset neural network model. The neural network model is determined based on the historical operating parameters and corresponding optimized compensation values ​​of the dual active full-bridge converter under different operating conditions.

[0042] The operation of the dual active full-bridge converter is controlled based on the external phase angle reference value and the external phase angle compensation value.

[0043] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0044] In the current control cycle, obtain the current operating parameters of the dual active full-bridge converter;

[0045] The reference value of the external phase shift angle of the dual active full-bridge converter is determined based on the operating parameters and the preset control algorithm;

[0046] The external phase shift compensation value of the dual active full-bridge converter is determined based on the operating parameters and the preset neural network model. The neural network model is determined based on the historical operating parameters and corresponding optimized compensation values ​​of the dual active full-bridge converter under different operating conditions.

[0047] The operation of the dual active full-bridge converter is controlled based on the external phase angle reference value and the external phase angle compensation value.

[0048] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0049] In the current control cycle, obtain the current operating parameters of the dual active full-bridge converter;

[0050] The reference value of the external phase shift angle of the dual active full-bridge converter is determined based on the operating parameters and the preset control algorithm;

[0051] The external phase shift compensation value of the dual active full-bridge converter is determined based on the operating parameters and the preset neural network model. The neural network model is determined based on the historical operating parameters and corresponding optimized compensation values ​​of the dual active full-bridge converter under different operating conditions.

[0052] The operation of the dual active full-bridge converter is controlled based on the external phase angle reference value and the external phase angle compensation value.

[0053] The aforementioned control method, device, system, computer-readable storage medium, and computer program product for a dual active full-bridge converter, in the current control cycle, calculates an external phase shift reference value based on the operating parameters of the dual active full-bridge converter using a control algorithm with good control stability under rated operating conditions. Furthermore, the current operating parameters of the dual active full-bridge converter are input into a neural network model to obtain an output external phase shift compensation value. Since the neural network model has learned optimized compensation values ​​under different operating conditions during the training phase, the combined control of the external phase shift reference value and the external phase shift compensation value enables stable operation of the dual active full-bridge converter under rated operating conditions and compensates for control deviations caused by changes in the operating conditions of the dual active full-bridge converter. Compared to using only a fixed control algorithm to control the operation of the dual active full-bridge converter, the control value obtained by combining the external phase shift reference value and the external phase shift compensation value can adapt to different operating conditions, thereby improving the performance of the dual active full-bridge converter under different operating conditions. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is an application environment diagram of the control method for a dual active full-bridge converter in one embodiment;

[0056] Figure 2 This is a flowchart illustrating the control method of a dual active full-bridge converter in one embodiment;

[0057] Figure 3 This is a detailed flowchart illustrating the steps of controlling the operation of a dual active full-bridge converter based on the external phase shift reference value and the external phase shift compensation value in one embodiment.

[0058] Figure 4 This is a schematic diagram of the control architecture of the controller in one embodiment;

[0059] Figure 5 This is a flowchart illustrating the control method for a dual active full-bridge converter in another embodiment;

[0060] Figure 6 This is a detailed flowchart illustrating the steps for determining the reference value of the outer phase shift angle of a dual active full-bridge converter based on operating parameters and a preset control algorithm in one embodiment.

[0061] Figure 7 This is a structural block diagram of the control device for a dual active full-bridge converter in one embodiment;

[0062] Figure 8 This is an internal structure diagram of a dual active full-bridge converter system in one embodiment. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0064] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0065] The control method for dual active full-bridge converters provided in this application embodiment can be applied to, for example... Figure 1 The illustrated dual active full-bridge converter system includes a dual active full-bridge converter 104 and a controller 102 connected to each other. The dual active full-bridge converter 104 receives a DC voltage, performs voltage conversion on the input DC voltage to obtain a transformed DC voltage, and outputs it. The controller 102, in each control cycle, acquires the current operating parameters of the dual active full-bridge converter 104, determines the reference value of the outer phase angle of the dual active full-bridge converter 104 based on the operating parameters and the control algorithm stored in the controller 102, and determines the compensation value of the outer phase angle of the dual active full-bridge converter 104 based on the operating parameters and the neural network model stored in the controller 102. Finally, the operation of the dual active full-bridge converter 104 is comprehensively controlled based on the reference value and the compensation value of the outer phase angle.

[0066] Specifically, the dual active full-bridge converter 104 includes a primary-side full-bridge circuit, a secondary-side full-bridge circuit, and a transformer. The transformer includes a primary winding and a secondary winding, with the primary-side full-bridge circuit connected to the primary winding and the secondary-side full-bridge circuit connected to the secondary winding. Both the primary-side and secondary-side full-bridge circuits include capacitors and at least four switching transistors. The primary-side full-bridge circuit may also include an inductor, which is used to limit sudden current changes in the primary-side full-bridge circuit and protect the power devices such as the switching transistors in the primary-side full-bridge circuit.

[0067] The primary-side full-bridge circuit receives a DC voltage, which is converted to an AC voltage by a switching transistor. The AC voltage is then transmitted to the transformer via an inductor. After being transformed by the transformer, the AC voltage is converted back to a DC voltage by a switching transistor in the secondary-side full-bridge circuit and then output to the power supply or load, thus completing the voltage conversion between DC and AC voltages.

[0068] The controller 102 can integrate a PWM (Pulse Width Modulation) module, or it can be connected to the dual active full-bridge converter 104 via an external PWM module. After obtaining the outward phase angle control value based on the outward phase angle reference value and the outward phase angle compensation value, the controller 102 inputs the outward phase angle control value to the PWM module. The PWM module generates two drive signals based on the outward phase angle control value, and the phase difference between the two drive signals is the outward phase angle control value. After generating the two drive signals, the PWM module outputs the two drive signals to the primary-side full-bridge circuit and the secondary-side full-bridge circuit, respectively, driving the switching transistors in the primary-side full-bridge circuit and the secondary-side full-bridge circuit, thereby changing the voltage conversion power of the DAB converter.

[0069] The controller 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, microcontrollers, cloud devices, and portable wearable devices, such as smartwatches and smart bracelets.

[0070] In one exemplary embodiment, such as Figure 2 As shown, a control method for a dual active full-bridge converter is provided, which is then applied to... Figure 1 Taking controller 102 as an example, the explanation includes the following steps 202 to 208. Wherein:

[0071] Step 202: In the current control cycle, obtain the current operating parameters of the dual active full-bridge converter.

[0072] The control cycle refers to the time interval between the controller performing one sampling and control operation on the dual active full-bridge converter. Operating parameters refer to physical quantities that characterize the current operating state of the converter, such as input voltage, inductor current, and output voltage.

[0073] In this embodiment, within each preset control cycle, the controller acquires the current operating parameters of the dual active full-bridge converter in real time. These operating parameters may include the input voltage, inductor current, and output voltage of the DAB converter. After acquiring these operating parameters, the controller temporarily stores them for use in the current control cycle. The controller can directly acquire the operating parameters of the DAB converter through hardware sampling circuitry, or it can estimate certain parameters that are difficult to measure directly through a state observer, such as inferring the load current from the input power and output voltage, and using this as supplementary operating parameters. Alternatively, provided that accuracy requirements are met, the controller can obtain the above operating parameters by reducing the sampling frequency or averaging multiple sample values ​​to filter out high-frequency noise interference.

[0074] Step 204: Determine the reference value of the external phase shift angle of the dual active full-bridge converter based on the operating parameters and the preset control algorithm.

[0075] The control algorithm is a method for determining the outer phase shift angle of the DAB converter based on its current operating parameters. It can be a deviation-based linear control algorithm, composed of proportional and integral actions, used to eliminate steady-state errors. Alternatively, it can be a lookup table method or a neural network model built based on a database of current operating parameters and the outer phase shift angle. The reference value for the outer phase shift angle is the fundamental outer phase shift angle value calculated by the proportional-integral controller based on the current operating parameters, used to control the power transfer of the dual active full-bridge converter; its range is between 0 and 180 degrees.

[0076] In this embodiment, the controller inputs the collected operating parameters into the control algorithm, which processes them to obtain the reference value of the outer phase angle for the current control cycle. This reference value aims to maintain the basic stability of the output voltage. The control algorithm can adopt a standard voltage and current cascaded dual closed-loop structure, or it can be simplified or adjusted according to the focus of the application scenario. For example, in situations where dynamic response requirements are not high, a single voltage closed-loop structure can be used, and the reference value of the outer phase angle can be calculated directly based on the voltage deviation. Alternatively, to suppress the inrush current, a limiting circuit can be added to the inner current loop to limit the maximum value of the inductor current setpoint.

[0077] In addition, during the initialization phase of the DAB converter startup, technicians can manually adjust the reference parameters of the PI control algorithm. The reference parameters include the proportional coefficient Kp and the integral coefficient Ki. These reference parameters can ensure the stable operation of the converter and maximize the basic stability of the system.

[0078] Specifically, when determining the baseline parameters of the PI control algorithm, a segmented testing method can be used. Under four typical operating conditions of the DAB converter—no load, light load, rated load, and overload—the proportional coefficient Kp and integral coefficient Ki are gradually adjusted, and the output voltage fluctuation and step response time of the DAB converter are recorded. The optimal parameters corresponding to the minimum output voltage fluctuation and the shortest step response time are selected as the baseline parameters for traditional PI control.

[0079] Step 206: Determine the external phase shift compensation value of the dual active full-bridge converter based on the operating parameters and the preset neural network model.

[0080] The neural network model is an artificial intelligence model that simulates the structure and function of a biological neural network. It can fit complex nonlinear mapping relationships by learning from historical data. The neural network model is determined based on the historical operating parameters of the dual active full-bridge converter under different operating conditions and the corresponding optimized compensation values. The external phase shift compensation value is an additional angle adjustment output by the neural network model to compensate for the offset of the control algorithm under complex operating conditions. Its value can be positive or negative and is used to superimpose on the reference value to optimize the stability of the DAB converter. The optimized compensation value is the ideal phase shift difference that enables the converter to achieve optimal performance indicators (such as minimum current stress and maximum efficiency) under specific operating conditions.

[0081] In this embodiment, the controller embeds a pre-trained neural network model. This model takes the operating parameters acquired in the current control cycle as input, performs forward computation, and outputs an outward phase angle compensation value. The training process of this neural network model is completed offline. Besides using a backpropagation neural network, other types of neural networks can also be used to achieve this function. For example, a radial basis function neural network can be used, which has a faster training speed and is suitable for applications with slightly higher real-time requirements. Alternatively, to reduce the computational power requirement, an extreme learning machine can be used, whose hidden layer parameters are randomly generated, requiring only the calculation of output weights, which greatly simplifies the training process and facilitates deployment on low-cost controllers.

[0082] Step 208: Control the operation of the dual active full-bridge converter based on the external phase angle reference value and the external phase angle compensation value.

[0083] In this embodiment, the controller algebraically superimposes the outer phase angle reference value output by the control algorithm with the outer phase angle compensation value output by the neural network model to calculate the final outer phase angle for the current control cycle. The controller then generates a corresponding pulse width modulation signal based on this final calculated outer phase angle and drives the switching of power electronic devices in the dual active full-bridge converter, thereby controlling the converter's power transmission. Under steady-state conditions, the compensation value output by the neural network model approaches zero, and the outer phase angle is mainly determined by the proportional-integral controller. Under transient or non-rated conditions, the outer phase angle compensation value actively intervenes to compensate for the outer phase angle reference value.

[0084] The aforementioned control method, device, system, computer-readable storage medium, and computer program product for dual active full-bridge converters, in the current control cycle, calculate an external phase shift reference value based on the operating parameters of the dual active full-bridge converter using a control algorithm with good control stability under rated operating conditions. Furthermore, the operating parameters of the current dual active full-bridge converter are input into a neural network model to obtain an output external phase shift compensation value. Since the neural network model has learned optimized compensation values ​​that meet user needs under different operating conditions during the training phase, the operation of the dual active full-bridge converter is comprehensively controlled by combining the external phase shift reference value with the external phase shift compensation value. This ensures stable operation of the dual active full-bridge converter under rated operating conditions and compensates for control deviations generated by the control algorithm when the operating conditions of the dual active full-bridge converter change. Ultimately, it achieves stable operation that meets the optimization objectives across the entire operating range, improving the stability of the DAB converter under wide operating conditions.

[0085] Specifically, in one exemplary embodiment, such as Figure 3 As shown, step 208 above includes steps 302 and 304, wherein:

[0086] Step 302: The outer phase angle control value is obtained by summing the outer phase angle reference value and the outer phase angle compensation value.

[0087] In this embodiment, after obtaining the reference value of the outer phase angle output by the control algorithm and the compensation value of the outer phase angle output by the neural network model, the controller adds the reference value and the compensation value to obtain the outer phase angle control value. In some embodiments, in addition to directly adding the two values, an enable switch or weighting coefficient can be introduced to control the degree of intervention of the compensation value. For example, when the DAB converter starts up or a serious fault occurs, the controller can set the coefficient of the compensation value to zero, shield the role of the neural network, and only use proportional-integral control to operate alone to ensure the absolute reliability of the system. Alternatively, depending on the different emphases of current stress and efficiency optimization, a gain coefficient less than 1 can be multiplied by the outer phase angle compensation value before superimposing the reference value and the compensation value to balance stability and optimize performance.

[0088] Step 304: Control the operation of the dual active full-bridge converter according to the external phase shift control value.

[0089] In this embodiment, the controller can integrate a PWM module or be connected to a dual active full-bridge converter via an external PWM module. After obtaining the outward phase shift control value based on the outward phase shift reference value and the outward phase shift compensation value, the controller inputs the outward phase shift control value to the PWM module. The PWM module generates two drive signals based on the outward phase shift control value, and the phase difference between the two drive signals is the outward phase shift control value. After generating the two drive signals, the PWM module outputs them to the primary-side full-bridge circuit and the secondary-side full-bridge circuit, respectively, driving the switching transistors in the primary-side and secondary-side full-bridge circuits to operate, thereby changing the voltage conversion power of the DAB converter.

[0090] Since the neural network model has learned the optimized compensation values ​​under different operating conditions during the training phase, the operation of the dual active full-bridge converter can be comprehensively controlled by combining the external phase angle reference value with the external phase angle compensation value. This can not only ensure the stable operation of the dual active full-bridge converter under rated operating conditions, but also compensate for the control deviation generated by the control algorithm when the operating conditions of the dual active full-bridge converter change.

[0091] like Figure 4 As shown, in each control cycle, after acquiring the operating parameters of the DAB converter, the controller inputs these parameters to the PI control section and the neural network section of the controller, respectively. In the PI control section, the operating parameters are processed using a PI control algorithm to obtain the external phase shift reference value α1. In the neural network section, the operating parameters are processed using a neural network model to obtain the external phase shift compensation value α2. Adding the external phase shift reference value α1 and the external phase shift compensation value α2 yields the final external phase shift α used as input to the main circuit of the DAB converter, thereby controlling the voltage conversion power of the DAB converter.

[0092] In one exemplary embodiment, such as Figure 5 As shown, the control method for the dual active full-bridge converter further includes steps 502 to 506, wherein:

[0093] Step 502: Obtain the historical operating parameters and corresponding optimized compensation values ​​of the dual active full-bridge converter under different operating conditions.

[0094] Historical operating parameters refer to data such as input voltage, inductor current, and output voltage recorded by the dual active full-bridge converter under various specific operating conditions during offline acquisition or simulation. The optimized compensation value refers to an outward phase angle difference obtained through experimentation or simulation for each specific set of historical operating parameters. When this difference is used to correct the proportional-integral controller output, it enables the converter's performance to reach the optimal level set by the technicians, such as minimizing current stress and maximizing voltage conversion efficiency. Performance index parameters refer to indicators that reflect the performance of the DAB converter during operation, such as current stress, voltage conversion efficiency, and load current. Preset optimization conditions can be set according to actual needs, such as minimizing current stress and maximizing transmission efficiency.

[0095] In this embodiment, the controller first needs to construct a dataset for training the neural network. To this end, the controller collects or simulates historical operating parameters of the dual active full-bridge converter under a series of typical operating conditions, including no-load, light-load, rated load, and overload. For each set of historical operating parameters collected, the controller determines a corresponding optimized compensation value through additional optimization algorithms or exhaustive testing. When determining this optimized compensation value, the DAB converter after applying the compensation needs to meet preset optimization conditions. These paired historical operating parameters and optimized compensation value data constitute the sample set for subsequent model training. Besides obtaining this data through experimental testing, the required samples can also be generated through offline simulation using the mathematical model of the dual active full-bridge converter. For example, all possible phase shift angles and operating points can be traversed in the simulation environment to directly calculate the ideal compensation value that minimizes current stress and maximizes efficiency under each operating condition. Alternatively, to reduce the sample size and improve training efficiency, an orthogonal experimental design method can be used to purposefully select a few representative key operating points for testing, rather than collecting data from all continuous operating conditions.

[0096] Step 504: Use historical operating parameters as samples and optimized compensation values ​​as labels to train the preset base model and obtain the intermediate model.

[0097] Here, the base model refers to an untrained neural network structure with initial random or preset weights and thresholds. The label refers to the target output value corresponding to the input sample in supervised learning, i.e., the optimized compensation value. The intermediate model refers to a stage model where the internal parameters (weights and thresholds) of the base model have been optimized to some extent after initial learning, but have not yet reached the final accuracy requirement.

[0098] In this embodiment, the controller uses the constructed dataset for supervised learning. Specifically, the controller feeds each set of historical running parameters as input data to a preset base model. The base model calculates a predicted compensation value based on its current internal parameters. The controller compares this predicted value with the corresponding label in the dataset, i.e., the true optimized compensation value. Based on the difference in the comparison results, the controller adjusts the internal weights and thresholds of the base model according to a preset learning algorithm (e.g., backpropagation), so that the base model's next prediction for the same sample is closer to the true label. The controller iterates through all samples in the dataset, thereby completing one round of training for the base model and obtaining an intermediate model with preliminary mapping capabilities. In addition to using gradient descent for training, online learning or mini-batch learning can also be used to update the model. For example, only one sample or a small batch of samples can be selected each time to calculate the error and update the model parameters. Although this method may have a more complex convergence path, it can reduce the computational cost of each iteration and obtain a usable intermediate model faster when the sample data volume is very large.

[0099] Step 506: Update the model parameters of the intermediate model according to the preset descent gradient, and determine the loss value of the intermediate model based on the preset loss function until the loss value is less than or equal to the preset loss threshold, then use the current intermediate model as a neural network model.

[0100] In this context, the descent gradient refers to a pre-set reduction amount. During each training cycle, this gradient is used to decrease the model parameters, thereby updating them. Model parameters include the weights and thresholds of all neurons in the neural network. The loss function is an evaluation function used to quantify whether the performance of the DAB transformer corresponding to the model's predicted values ​​meets the optimization conditions. The loss threshold is a preset small value used as a termination condition to determine whether the model training has reached the required accuracy.

[0101] In this embodiment, after obtaining the intermediate model, the controller uses a loss function to comprehensively evaluate the model's performance. This loss function can be determined by optimization conditions set for one objective, or it can be determined by optimization conditions set for multiple objectives simultaneously. When determining the loss function for optimization conditions set for multiple objectives, the loss function can simultaneously quantify the prediction errors of the DAB model corresponding to the predicted output of the neural network model on three objectives: current stress, transmission efficiency, and output voltage.

[0102] Specifically, the loss function can be a multi-objective weighted loss function, which is as follows:

[0103]

[0104] in, The current stress loss term uses the mean square error between the actual current stress and the optimal current stress, i.e. , The actual current stress corresponding to the i-th training sample; For the efficiency loss term, the mean square error between the optimal efficiency and the actual efficiency is used, i.e. , The actual transmission efficiency of the i-th sample; To compensate for the phase shift angle deviation loss term, the mean square error between the actual output compensation value and the optimal compensation value of the neural network is used, i.e. , V represents the actual output voltage of the converter corresponding to the i-th sample. ref,i This is the reference value for the output voltage under the i-th sample operating condition, used to ensure output voltage stability; This is a weighting coefficient, which can be flexibly adjusted according to the actual control priority.

[0105] The controller calculates the loss value of the current intermediate model based on the loss function. If the loss value is greater than the loss threshold, all model parameters are updated according to the descent gradient. The controller repeats this iterative loop of calculating loss, comparing loss, and updating model parameters. After each iteration, the controller recalculates the loss value and compares it with a preset loss threshold (e.g., 0.001). When the loss value is reduced to less than or equal to the loss threshold, it indicates that the model's prediction accuracy has met the design requirements. At this point, the controller stops iteratively updating and uses the current intermediate model as the final neural network model that can be deployed in the controller. In addition to using a fixed learning rate for gradient descent, an optimization algorithm with an adaptive learning rate can also be used to update the model parameters; no further limitations are made here.

[0106] Through the above steps, a neural network model can be obtained that can accurately adapt to the nonlinear characteristics of the dual active full-bridge converter and output the optimal compensation value in the entire operating range. This model can compensate for the offset of the PI control algorithm in the entire operating range and improve the stability of the DAB converter.

[0107] In an exemplary embodiment, after step 208 described above, the control method for the dual active full-bridge converter further includes: obtaining the current stress of the inductor in the dual active full-bridge converter and the voltage conversion efficiency of the dual active full-bridge converter; updating the neural network model based on the current stress when the rise of the current stress is within a preset first range; and / or updating the neural network model based on the voltage conversion efficiency when the decrease of the voltage conversion efficiency is within a preset second range; the updated neural network model is used as the preset neural network model in the next control cycle.

[0108] The increase in current stress refers to the percentage increase in the currently measured current stress relative to the baseline value under rated operating conditions or the previous stable operating condition. The preset first range is a small numerical interval, for example, greater than 5% but less than or equal to 10%, used to define a slight but noteworthy degradation in current stress. The decrease in voltage conversion efficiency refers to the percentage decrease in the currently measured transmission efficiency relative to the optimal or rated efficiency. The preset second range is also a small numerical interval, for example, greater than 1% but less than or equal to 3%, used to define a state where efficiency has slightly decreased but not to the point of triggering emergency protection.

[0109] In this embodiment, the controller continuously acquires the current stress and voltage conversion efficiency of the inductors in the DAB converter during the operation of the DAB converter. In some embodiments, when the controller detects that the increase in current stress is within a preset first range, the controller determines that the converter may be experiencing performance degradation. At this time, the controller performs a lightweight update of the neural network model online. The controller uses the currently acquired operating parameters and the measured current stress as a new set of incremental data, and adjusts the weights and thresholds of the currently used neural network model one or a finite number of times according to preset rules (e.g., using gradient descent). After the adjustment is completed, the controller saves the neural network model with updated parameters and uses it as the neural network model on which the outward phase angle compensation value is calculated in the next control cycle.

[0110] In other embodiments, when the controller detects that the decrease in voltage conversion efficiency is within a preset second range, the controller determines that the converter may be experiencing performance degradation. At this time, the controller uses the currently acquired operating parameters and the measured voltage conversion efficiency as a new set of incremental data, and adjusts the weights and thresholds of the currently used neural network model one or a limited number of times according to preset rules. After fine-tuning, the controller saves the neural network model with updated parameters and uses it as the neural network model for calculating the outward phase angle compensation value in the next control cycle.

[0111] In some embodiments, when the controller detects that the increase in current stress is within a preset first range and the decrease in voltage conversion efficiency is within a preset second range, the controller determines that the converter may be experiencing performance degradation. At this time, the controller uses the currently acquired operating parameters, along with the measured operating parameters and voltage conversion efficiency, as a new set of incremental data. It then adjusts the weights and thresholds of the currently used neural network model one or a limited number of times according to preset rules. After the adjustment is complete, the controller saves the neural network model with updated parameters and uses it as the neural network model for calculating the outward phase angle compensation value in the next control cycle.

[0112] In the above embodiments, besides using online gradient descent to adjust the neural network, other online learning strategies can also be used to update the model. For example, the controller can use the difference between the outward phase angle when performance deteriorates and the current optimal outward phase angle as the control deviation, and store several sets of operating parameters that recently caused performance degradation and their corresponding control deviations. When any of the update conditions in the above embodiments are met, the stored multiple sets of data are used to perform a quick retraining or parameter interpolation correction on the neural network model. Alternatively, to simplify the computation of online updates, an offset coefficient can be directly compensated based on a basic outward phase angle compensation value output by the neural network model, and this offset coefficient can be superimposed with the output of the neural network as the final outward phase angle compensation value.

[0113] By introducing an online adaptive update mechanism, when the DAB converter experiences a slight performance degradation due to slow operating condition drift or device parameter aging, the controller can promptly trigger incremental updates to the neural network model, enabling the neural network model to quickly adapt to the new operating environment and maintain current stress and transmission efficiency at near-optimal levels.

[0114] In an exemplary embodiment, after step 208 described above, the control method for the dual active full-bridge converter further includes: acquiring the input voltage of the dual active full-bridge converter and the soft-switching state of the power devices in the dual active full-bridge converter; and controlling the operation of the dual active full-bridge converter based on the external phase angle reference value when the input voltage is within a preset fault range and / or the soft-switching state is soft-switching failure.

[0115] The preset fault range refers to the interval where the input voltage deviates from the rated value to a certain extent but has not yet triggered the power-off protection, such as exceeding the rated voltage by ±5% to ±10%. Soft-switching state refers to the situation where the power switching devices in the dual active full-bridge converter achieve zero-voltage turn-on or zero-current turn-off. Soft-switching failure indicates that the switching devices failed to complete the turn-on or turn-off under ideal conditions (i.e., zero voltage), resulting in increased switching losses and aggravated electromagnetic interference.

[0116] In this embodiment, the controller monitors the input voltage and soft-switching status of the power devices in the DAB converter in real time during its operation. In some embodiments, when the controller detects that the input voltage is within a preset fault range, it determines that the converter is currently experiencing a fault or that the operating condition deviates significantly from the design point. In this case, since the outward phase angle compensation value output by the neural network dynamic compensation module is obtained based on normal operating conditions or offline training data, it may output incorrect control quantities under fault or extreme conditions, which may exacerbate system instability. Therefore, the controller immediately performs fault-tolerant operation, cutting off the output of the neural network dynamic compensation module, setting the outward phase angle compensation value output by the neural network module to zero, and using only the outward phase angle reference value output by the control algorithm to control the operation of the dual active full-bridge converter. At the same time, the controller sends a fault warning signal to the system host computer or user, indicating that there is an abnormal input voltage or soft-switching failure, so that manual intervention can be carried out for troubleshooting.

[0117] In other embodiments, when the controller detects a failure in the soft-switching state of the power devices, it determines that the converter is currently experiencing a fault or its operating condition deviates significantly from the design point. In this case, since the outward phase angle compensation value output by the neural network dynamic compensation module is obtained based on normal operating conditions or offline training data, it may output incorrect control quantities under fault or extreme conditions, thereby exacerbating system instability. Therefore, the controller immediately performs fault-tolerant operation, cutting off the output of the neural network dynamic compensation module, setting the outward phase angle compensation value output by the neural network module to zero, and using only the outward phase angle reference value output by the control algorithm to control the operation of the dual active full-bridge converter. Simultaneously, the controller sends a fault warning signal to the system host computer or user, indicating the presence of abnormal input voltage or soft-switching failure, so that manual intervention can be performed for troubleshooting.

[0118] In some embodiments, when the controller detects that the input voltage is within a preset fault range and that the soft-switching state of the power devices has failed, the controller determines that the converter is currently experiencing a fault or that the operating condition has significantly deviated from the design point. In this case, since the outward phase angle compensation value output by the neural network dynamic compensation module is obtained based on normal operating conditions or offline training data, it may output incorrect control quantities under fault or extreme conditions, which may exacerbate system instability. Therefore, the controller immediately performs fault-tolerant operation, cutting off the output of the neural network dynamic compensation module, setting the outward phase angle compensation value output by the neural network module to zero, and using only the outward phase angle reference value output by the control algorithm to control the operation of the dual active full-bridge converter. At the same time, the controller sends a fault warning signal to the system host computer or user, indicating that there is an abnormal input voltage or soft-switching failure, so that manual intervention can be carried out for troubleshooting.

[0119] The above steps, in the event of a DAB converter malfunction or a significant deviation from its operating conditions, employ a basic PI control algorithm to maintain the basic operation of the DAB converter while simultaneously alerting staff to conduct manual troubleshooting. This reduces the risk of the DAB converter continuing to operate in a faulty state and improves the stability of the DAB converter.

[0120] In an exemplary embodiment, after step 208 above, the control method for the dual active full-bridge converter further includes: acquiring the device temperature of the power devices in the dual active full-bridge converter and the load operating status of the load connected to the dual active full-bridge converter; and controlling the dual active full-bridge converter to power off when the load operating status is a load short circuit and / or the device temperature is greater than or equal to a preset temperature threshold.

[0121] Here, device temperature refers to the junction temperature or case temperature of the power switching devices in the dual active full-bridge converter. Load operating status refers to the operating condition of the electrical equipment connected to the output of the dual active full-bridge converter; a load short circuit indicates an abnormal electrical connection with low or zero impedance at the output. The preset temperature threshold is the highest permissible temperature limit to ensure the safe operation of power devices, for example, 85 degrees Celsius. Exceeding this value may cause device damage or a sharp decline in performance.

[0122] In this embodiment, the controller continuously monitors the device temperature of the power devices in the DAB converter and the status of the load connected to the DAB converter during operation. In some embodiments, when the controller detects that the load is in a short-circuit state, the controller determines that the converter has encountered a serious fault. If such a fault is not dealt with immediately, it will cause irreversible physical damage to the converter itself, the load equipment, and even the equipment itself. Therefore, the controller will no longer execute any form of online adjustment or degraded operation strategy, but will immediately cut off the power supply circuit of the dual active full-bridge converter, stop the switching action of all power switching devices, and put the converter into a complete power-off shutdown protection state. At the same time, the controller sends an emergency fault warning signal to the system host computer, local display interface, or remote monitoring center, and records the fault type and occurrence time for subsequent fault investigation and equipment maintenance.

[0123] In other embodiments, when the controller detects that the temperature of a power device has reached or exceeded a preset safe temperature threshold, the controller determines that the converter has encountered a serious fault. If such a fault is not addressed immediately, it will cause irreversible physical damage to the converter itself, the load equipment, and even the system itself. Therefore, the controller no longer executes any form of online adjustment or degraded operation strategy, but directly triggers the highest level of protection operation, immediately cutting off the power supply circuit of the dual active full-bridge converter, stopping the switching action of all power switching devices, and putting the converter into a complete power-off shutdown protection state. Simultaneously, the controller sends an emergency fault warning signal to the system's host computer, local display interface, or remote monitoring center, recording the fault type and occurrence time for subsequent fault investigation and equipment maintenance.

[0124] In some embodiments, when the controller detects that the load is operating in a short-circuit state and the device temperature of the power devices reaches or exceeds a preset safe temperature threshold, the controller determines that the converter has encountered a serious fault. If such a fault is not addressed immediately, it will cause irreversible physical damage to the converter itself, the load equipment, and even the system itself. Therefore, the controller no longer executes any form of online adjustment or degraded operation strategy, but directly triggers the highest level of protection operation, immediately cutting off the power supply circuit of the dual active full-bridge converter, stopping the switching action of all power switching devices, and putting the converter into a completely power-off shutdown protection state. Simultaneously, the controller sends an emergency fault warning signal to the system's host computer, local display interface, or remote monitoring center, recording the fault type and occurrence time for subsequent fault investigation and equipment maintenance.

[0125] The above steps facilitate timely identification of serious abnormalities in the DAB converter and enable power-off protection for the DAB converter, reducing the risk of damage to the DAB converter and its connected load devices due to continuous operation under serious faults.

[0126] In one exemplary embodiment, such as Figure 6 As shown, the operating parameters include the output voltage and inductor current values; the control algorithm includes an outer voltage loop algorithm and an inner current loop algorithm; step 204 above includes steps 602 to 608, wherein:

[0127] Step 602: Determine the voltage deviation value based on the output voltage and the preset reference voltage.

[0128] Here, output voltage refers to the actual DC voltage value output by the dual active full-bridge converter. The preset reference voltage is a target value for the desired output voltage, set by the user according to application requirements. Voltage deviation refers to the difference between the actual output voltage and the reference voltage.

[0129] In this embodiment, the controller acquires the actual output voltage of the dual active full-bridge converter in real time and subtracts it from the internally stored preset reference voltage to calculate the current voltage deviation value. This voltage deviation value reflects the degree and direction of the output voltage deviating from the desired target: when the actual output voltage is lower than the reference voltage, the voltage deviation value is positive, indicating that the power transfer of the DAB converter needs to be increased to improve the output voltage; when the actual output voltage is higher than the reference voltage, the voltage deviation value is negative, indicating that the power transfer needs to be reduced to decrease the output voltage. The controller uses this voltage deviation value as input for subsequent voltage outer loop algorithm processing.

[0130] Step 604: Determine the inductor current setpoint of the dual active full-bridge converter based on the voltage deviation value and the voltage outer loop algorithm.

[0131] The voltage outer loop algorithm refers to a calculation method that uses proportional-integral control logic to adjust the controlled variable of the outer loop. In this scheme, the voltage outer loop uses the output voltage as the controlled object. The inductor current setpoint refers to the desired current value calculated by the voltage outer loop controller based on the voltage deviation. This value will serve as the tracking target of the inner loop and be used to indirectly control the output voltage.

[0132] In this embodiment, the controller inputs the calculated voltage deviation value into the voltage outer loop algorithm. The voltage outer loop algorithm processes the voltage deviation value according to preset proportional and integral coefficients, and the output of the voltage outer loop algorithm is the inductor current setpoint. This setpoint represents the target value that the inductor current should reach in order to make the output voltage approach the reference voltage.

[0133] Step 606: Determine the current deviation value based on the inductor current setpoint and the inductor current value.

[0134] The inductor current value refers to the actual current value flowing through the inductor in the dual active full-bridge converter, which is collected in real time by the controller. The current deviation value refers to the difference between the inductor current setpoint and the actual inductor current value.

[0135] In this embodiment, the controller compares the determined inductor current setpoint with the real-time acquired actual inductor current value, and calculates the current deviation value through subtraction. This current deviation value reflects the degree of error in the actual inductor current tracking the desired current: when the actual inductor current is lower than the setpoint, the current deviation value is positive, indicating that the outer phase angle needs to be increased to increase the current; when the actual inductor current is higher than the setpoint, the current deviation value is negative, indicating that the outer phase angle needs to be decreased to decrease the current. The controller uses this current deviation value as input for subsequent current inner loop algorithm processing.

[0136] Step 608: Determine the reference value of the outer phase shift angle of the dual active full-bridge converter based on the current deviation value and the current inner loop algorithm.

[0137] The current inner loop algorithm refers to a calculation method that uses proportional-integral control logic to adjust the controlled variable in the inner loop. In this scheme, the inductor current is the controlled object in the current inner loop. The outer phase shift reference value is the final output value calculated by the current inner loop proportional-integral controller based on the current deviation, which is used to directly control the power transmission angle between the primary and secondary sides of the dual active full-bridge converter.

[0138] In this embodiment, the controller inputs the calculated current deviation value into the current inner loop algorithm. The output of the current inner loop algorithm is the reference value of the outer phase shift angle of the dual active full-bridge converter. This reference value ranges from 0 degrees to 180 degrees and directly determines the magnitude and direction of the converter's transmitted power. The controller saves this reference value for subsequent superposition with the outer phase shift angle compensation value.

[0139] Compared with the traditional single-voltage-loop control algorithm, the added current inner loop in this dual-closed-loop PI control algorithm can instantly sense and suppress the inductor current surge and overshoot, which is beneficial to improving the dynamic response performance and control quality of the converter in transient processes such as load changes.

[0140] To better understand the above embodiments, an optional embodiment will be explained in detail below.

[0141] Before the DAB converter is initially powered on, the controller first needs to initialize its PI control algorithm and neural network model. First, using a segmented testing method, under four typical operating conditions—no load, light load, rated load, and overload—the proportional coefficient Kp and integral coefficient Ki are gradually adjusted, and the output voltage fluctuation and step response time of the converter are recorded. The optimal parameters with the minimum output voltage fluctuation and shortest step response time are selected as the baseline parameters for PI control. Simultaneously, historical operating parameters of the DAB converter under different operating conditions, or historical operating parameters obtained from simulation, and the corresponding optimized compensation values ​​are used as samples to train the neural network model. A multi-objective weighted loss function is used as a quantitative indicator to measure whether the compensation value output by the neural network model meets the optimization conditions. Gradient descent is used to iteratively optimize the weights and thresholds of the neural network until the loss value calculated by the neural network model under the multi-objective weighted loss function is less than or equal to 0.001, at which point training stops, and the current neural network model is stored.

[0142] After initializing the PI control algorithm and neural network model, the DAB converter powers on and runs. In each control cycle, the controller acquires the current operating parameters of the DAB converter and inputs these parameters into the PI control algorithm and neural network model respectively, obtaining the external phase shift reference value and external phase shift compensation value. The external phase shift reference value and external phase shift compensation value are used to obtain the final external phase shift angle, which is then used to control the DAB converter. During the operation of the DAB converter, the controller continuously acquires the performance parameters of the DAB converter. If the increase in current stress is within a first range, or the decrease in voltage conversion efficiency is within a second range, the controller recognizes that the currently used neural network model may need optimization. At this time, the controller updates the neural network model based on a pre-set optimization method and uses the updated neural network model as the neural network model for subsequent control cycles. If the controller detects that the input voltage is within a fault range, or that the soft-switching state of the switching devices within the DAB converter has failed, the controller determines that the DAB converter is in an abnormal state, sets the output of the neural network model to zero, and controls the DAB converter's operation using the external phase shift reference value. If the controller detects a short circuit in the load connected to the DAB converter, or if the temperature of the switching devices within the DAB converter exceeds or equals a temperature threshold, it determines that the DAB converter has a serious fault and immediately cuts off power to the DAB converter to ensure its safety. If the controller detects that all performance parameters of the DAB converter are normal, it will not execute feedback control operations until the next control cycle.

[0143] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0144] Based on the same inventive concept, this application also provides a control device for a dual active full-bridge converter for implementing the control method of the dual active full-bridge converter described above. The solution provided by this device is similar to the implementation described in the above method. Therefore, the specific limitations in one or more embodiments of the control device for a dual active full-bridge converter provided below can be found in the limitations of the control method for the dual active full-bridge converter described above, and will not be repeated here.

[0145] In one exemplary embodiment, such as Figure 7 As shown, a control device for a dual active full-bridge converter is provided, comprising:

[0146] The data acquisition module 701 is used to acquire the current operating parameters of the dual active full-bridge converter in the current control cycle;

[0147] The reference control module 702 is used to determine the reference value of the external phase shift angle of the dual active full-bridge converter based on the operating parameters and the preset control algorithm.

[0148] The compensation module 703 is used to determine the external phase shift compensation value of the dual active full-bridge converter based on the operating parameters and the preset neural network model. The neural network model is determined based on the historical operating parameters of the dual active full-bridge converter under different operating conditions and the corresponding optimized compensation values.

[0149] The control module 704 is used to control the operation of the dual active full-bridge converter based on the external phase angle reference value and the external phase angle compensation value.

[0150] In one embodiment, the control device for the dual active full-bridge converter further includes a modeling module, used to acquire historical operating parameters and corresponding optimized compensation values ​​of the dual active full-bridge converter under different operating conditions, wherein the performance index parameters of the dual active full-bridge converter corresponding to the optimized compensation values ​​meet preset optimization conditions; the historical operating parameters are used as samples and the optimized compensation values ​​are used as labels to train a preset base model to obtain an intermediate model; the model parameters of the intermediate model are updated according to a preset descent gradient, and the loss value of the intermediate model is determined based on a preset loss function until the loss value is less than or equal to a preset loss threshold, at which point the current intermediate model is used as a neural network model.

[0151] In one embodiment, the control module 704 is further configured to obtain an outer phase angle control value by summing the outer phase angle reference value and the outer phase angle compensation value; and to control the operation of the dual active full-bridge converter according to the outer phase angle control value.

[0152] In one embodiment, the control device for the dual active full-bridge converter further includes a feedback module for acquiring the current stress of the inductor and the voltage conversion efficiency of the dual active full-bridge converter; updating the neural network model based on the current stress when the rise of the current stress is within a preset first range; and / or updating the neural network model based on the voltage conversion efficiency when the decrease of the voltage conversion efficiency is within a preset second range; the updated neural network model serves as the preset neural network model for the next control cycle.

[0153] In one embodiment, the feedback module is further configured to acquire the input voltage of the dual active full-bridge converter and the soft-switching state of the power devices in the dual active full-bridge converter; and control the operation of the dual active full-bridge converter based on the external phase angle reference value when the input voltage is within a preset fault range and / or the soft-switching state is soft-switching failure.

[0154] In one embodiment, the feedback module is further configured to acquire the device temperature of the power devices in the dual active full-bridge converter and the load operating status of the load connected to the dual active full-bridge converter; and control the dual active full-bridge converter to power off when the load operating status is a load short circuit and / or the device temperature is greater than or equal to a preset temperature threshold.

[0155] In one embodiment, the operating parameters include the output voltage and the inductor current value; the control algorithm includes a voltage outer loop algorithm and a current inner loop algorithm; the reference control module 702 is further configured to determine the voltage deviation value based on the output voltage and a preset reference voltage; determine the inductor current setpoint of the dual active full-bridge converter based on the voltage deviation value and the voltage outer loop algorithm; determine the current deviation value based on the inductor current setpoint and the inductor current value; and determine the outer phase shift reference value of the dual active full-bridge converter based on the current deviation value and the current inner loop algorithm.

[0156] Each module in the control device of the aforementioned dual active full-bridge converter can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the controller in hardware form or independent of it, or stored in the memory of the controller in software form, so that the processor can call and execute the corresponding operations of each module.

[0157] In one exemplary embodiment, a dual active full-bridge converter system is provided, including dual active full-bridge converters and a controller interconnected. In some embodiments, the internal structure of the controller can be as follows: Figure 8 As shown, the controller includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs in the non-volatile storage media to run. The database stores the controller's control data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a control method for a dual active full-bridge converter.

[0158] Those skilled in the art will understand that Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the controller to which the present application is applied. A specific controller may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0159] In an exemplary embodiment, a dual active full-bridge converter system is provided, including dual active full-bridge converters and a controller connected to each other; the controller is used to control the operation of the dual active full-bridge converter according to the steps in the above-described control method embodiment of the dual active full-bridge converter, wherein during the operation of the dual active full-bridge converter, the input DC voltage is converted to obtain a transformed DC voltage and output.

[0160] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps in the control method embodiment of the dual active full-bridge converter described above.

[0161] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the control method embodiment for the dual active full-bridge converter described above.

[0162] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0163] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0164] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A control method for a dual active full-bridge converter, characterized in that, include: In the current control cycle, obtain the current operating parameters of the dual active full-bridge converter; The reference value of the external phase shift angle of the dual active full-bridge converter is determined based on the operating parameters and the preset control algorithm. The external phase shift compensation value of the dual active full-bridge converter is determined based on the operating parameters and the preset neural network model, wherein the neural network model is determined based on the historical operating parameters and corresponding optimized compensation values ​​of the dual active full-bridge converter under different operating conditions. The operation of the dual active full-bridge converter is controlled based on the external phase angle reference value and the external phase angle compensation value.

2. The method according to claim 1, characterized in that, The method further includes: The historical operating parameters and corresponding optimized compensation values ​​of the dual active full-bridge converter under different operating conditions are obtained, and the performance index parameters of the dual active full-bridge converter corresponding to the optimized compensation values ​​meet the preset optimization conditions. Using the historical operating parameters as samples and the optimized compensation values ​​as labels, a preset base model is trained to obtain an intermediate model; The intermediate model parameters are updated according to the preset descent gradient, and the loss value of the intermediate model is determined based on the preset loss function. When the loss value is less than or equal to the preset loss threshold, the current intermediate model is used as the neural network model.

3. The method according to claim 1, characterized in that, The control of the dual active full-bridge converter based on the external phase shift reference value and the external phase shift compensation value includes: The outer phase angle control value is obtained by summing the outer phase angle reference value and the outer phase angle compensation value; The operation of the dual active full-bridge converter is controlled according to the external phase shift control value.

4. The method according to claim 1, characterized in that, After controlling the operation of the dual active full-bridge converter based on the external phase shift reference value and the external phase shift compensation value, the method further includes: Obtain the current stress of the inductor in the dual active full-bridge converter and the voltage conversion efficiency of the dual active full-bridge converter; If the increase in current stress is within a preset first range, the neural network model is updated based on the current stress; and / or, if the decrease in voltage conversion efficiency is within a preset second range, the neural network model is updated based on the voltage conversion efficiency; the updated neural network model is used as the preset neural network model in the next control cycle.

5. The method according to claim 1, characterized in that, After controlling the operation of the dual active full-bridge converter based on the external phase shift reference value and the external phase shift compensation value, the method further includes: Obtain the input voltage of the dual active full-bridge converter and the soft-switching state of the power devices in the dual active full-bridge converter; When the input voltage is within a preset fault range, and / or the soft-switching state is a soft-switching failure, the dual active full-bridge converter is controlled to operate based on the external phase angle reference value.

6. The method according to claim 1, characterized in that, After controlling the operation of the dual active full-bridge converter based on the external phase shift reference value and the external phase shift compensation value, the method further includes: The device temperature of the power devices in the dual active full-bridge converter and the load operating status of the load connected to the dual active full-bridge converter are obtained. When the load is in a short-circuit state and / or the device temperature is greater than or equal to a preset temperature threshold, the dual active full-bridge converter is powered off.

7. The method according to any one of claims 1-6, characterized in that, The operating parameters include output voltage and inductor current values; the control algorithm includes an outer voltage loop algorithm and an inner current loop algorithm; determining the outer phase shift reference value of the dual active full-bridge converter based on the operating parameters and the preset control algorithm includes: The voltage deviation value is determined based on the output voltage and the preset reference voltage; The inductor current setpoint of the dual active full-bridge converter is determined based on the voltage deviation value and the voltage outer loop algorithm. The current deviation value is determined based on the given inductor current value and the inductor current value. The reference value of the outer phase shift angle of the dual active full-bridge converter is determined based on the current deviation value and the current inner loop algorithm.

8. A control device for a dual active full-bridge converter, characterized in that, include: The data acquisition module is used to acquire the current operating parameters of the dual active full-bridge converter during the current control cycle. The reference control module is used to determine the reference value of the external phase shift angle of the dual active full-bridge converter based on the operating parameters and the preset control algorithm. The compensation module is used to determine the external phase shift compensation value of the dual active full-bridge converter based on the operating parameters and the preset neural network model, wherein the neural network model is determined based on the historical operating parameters of the dual active full-bridge converter under different operating conditions and the corresponding optimized compensation values; The control module is used to control the operation of the dual active full-bridge converter based on the external phase angle reference value and the external phase angle compensation value.

9. A dual active full-bridge converter system, characterized in that, This includes interconnected dual active full-bridge converters and controllers; The controller is used to control the operation of the dual active full-bridge converter according to the steps of the method according to any one of claims 1 to 7. During the operation of the dual active full-bridge converter, the input DC voltage is converted to obtain a transformed DC voltage and then output.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images