A phase-shift modulation method for a dual active bridge DC-DC converter

By employing Latin hypercube sampling and a multilayer perceptron neural network model in a dual active bridge DC-DC converter, the permeability is dynamically adjusted, solving the problem of operating parameter mismatch caused by core nonlinearity and improving the converter's efficiency and current stress management.

CN117792011BActive Publication Date: 2026-11-17ZHENGZHOU UNIVERSITY OF LIGHT INDUSTRY
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202310738462.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-06-21
Publication Date
2026-11-17
Estimated Expiration
2043-06-21

AI Technical Summary

Technical Problem

In the existing technology, the dual active bridge DC-DC converter ignores the nonlinear factors of the high-frequency transformer core during the process of increasing capacity, resulting in mismatch of operating parameters, reduced efficiency and increased current stress.

Method used

The Latin hypercube sampling method is used to randomly sample design variables. The inductance is calculated by combining the finite element method and the field-circuit coupling method. A surrogate model is constructed using a multilayer perceptron artificial neural network to calculate the inductance in real time and optimize the phase shift, and a single-phase-shift modulation strategy is formulated.

Benefits of technology

By considering the nonlinear factors of the high-frequency transformer core and dynamically adjusting the permeability, the power transmission efficiency and current stress of the large-capacity DC-DC converter are improved, and more optimized phase-shift modulation is achieved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117792011B_ABST
    Figure CN117792011B_ABST
Patent Text Reader

Abstract

This invention provides a phase-shift modulation method for a dual active bridge DC-DC converter, belonging to the field of power electronic switch control technology. The method includes the following steps: randomly sampling the design variables primary voltage, secondary voltage, and phase shift ratio within the design variable sample space to obtain multiple different variable combinations; solving for the initial equivalent inductance based on the initial permeability and control equations; calculating the primary and secondary current values ​​at the initial time step and using these values ​​as the current excitation for the high-frequency transformer to obtain the permeability for the next time step; iterating the obtained permeability using the field-circuit coupling method to calculate the equivalent inductance of the primary and secondary windings at different time steps, obtaining datasets of the primary and secondary inductances corresponding to different time steps, and constructing an artificial neural network surrogate model; and using the model to calculate the primary and secondary inductances of the high-frequency transformer in real time at different times to obtain the instantaneous transmission power of the high-frequency transformer and determine the real-time optimal phase shift ratio.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of power electronic switch control technology, specifically relating to a phase-shift modulation method for a dual active bridge DC-DC converter. Background Technology

[0002] DC-DC converters are of great significance for high-density locomotive traction power supply, electric vehicle charging stations, new energy DC aggregation, and integrated DC transmission and distribution construction. Among them, the dual active bridge DC-DC converter with high-frequency transformer has advantages such as high voltage transfer ratio, high power density, easy soft switching, high modularity, and fast dynamic response, and is considered to be the most suitable circuit topology for medium- and high-voltage, large-capacity DC-DC conversion.

[0003] Currently, most research treats high-frequency transformers as equivalent to simple linear inductors, often neglecting the influence of core nonlinearity. However, with the increasing capacity of DC-DC converters, the core operates in a higher magnetic flux density range, significantly enhancing core nonlinearity and magnetic saturation. Frequent operation of the high-frequency transformer core in the saturation magnetic flux density range leads to nonlinearity in leakage inductance, resulting in mismatch of DAB converter operating parameters, reduced DAB converter operating efficiency, and significant current stress. Summary of the Invention

[0004] To improve the operating efficiency and reduce current stress of a dual active bridge high-capacity DC-DC converter, this invention provides a phase-shifting modulation method for a dual active bridge DC-DC converter, comprising the following steps:

[0005] The Latin hypercube sampling method is used to randomly sample the design variables of the high-frequency transformer of the converter, namely primary voltage V1, secondary voltage V2 and shift ratio D, in the design variable sample space to obtain multiple different combinations of design variables V1, V2 and D.

[0006] The initial equivalent inductance of the primary winding and the secondary winding are solved based on the initial permeability of the high-frequency transformer and the finite element control equations.

[0007] The initial time step primary current value and secondary current value are calculated based on the initial equivalent inductance of the primary and secondary windings, and the primary and secondary current values ​​are used as the current excitation of the high-frequency transformer to obtain the permeability of the next time step.

[0008] The field-circuit coupling method is used to iterate the permeability of the next time step in order to calculate the equivalent inductance of the primary and secondary windings at different time steps, thereby obtaining a dataset of primary and secondary inductances corresponding to multiple different combinations of variables at different time steps.

[0009] The dataset is input into a multilayer perceptron artificial neural network to construct an artificial neural network surrogate model of the variable combination and the primary and secondary inductances;

[0010] The artificial neural network proxy model is used to calculate the primary and secondary inductances of the high-frequency transformer at different times in real time, and the instantaneous transmission power of the high-frequency transformer based on the equivalent circuit of the dual active bridge DC-DC converter is obtained based on the primary and secondary inductances of the high-frequency transformer at different times.

[0011] The real-time optimal shift ratio D of the high-frequency transformer is determined based on the instantaneous transmission power. opt .

[0012] Preferably, the method for obtaining the combination of multiple variables is as follows:

[0013] Sampling is performed in a 3-dimensional variable space consisting of the primary voltage V1, secondary voltage V2, and shift ratio.

[0014] Divide each dimension into m non-overlapping intervals of equal length;

[0015] Randomly select a point in each interval of each dimension;

[0016] The randomly selected points are combined into a vector, forming multiple combinations of variables.

[0017] Preferably, the initial permeability is determined based on the basic magnetization curve, and the initial permeability is the slope of the tangent line to the basic magnetization curve at the origin.

[0018] Preferably, the governing equation is:

[0019]

[0020]

[0021]

[0022] Where μ is the permeability, A is the vector magnetic potential, σ is the conductivity, φ is the scalar potential, and J is the current density.

[0023] Preferably, the initial equivalent inductance of the primary winding and the secondary winding is calculated using the following formula:

[0024]

[0025] In the formula, W is the magnetic field energy, and I... pt / st Let be the current in the primary and secondary windings at time t;

[0026] The formula for calculating the magnetic field energy W is:

[0027]

[0028] In the formula, V represents the calculation region.

[0029] Preferably, the formulas for calculating the primary current value and the secondary current value are as follows:

[0030]

[0031] In the formula, i L Let i be the current flowing through inductor L. p i s These represent the primary and secondary current values, respectively; N is the turns ratio; and V1 and V2 are the primary and secondary voltages, respectively.

[0032] Preferably, the process of constructing the artificial neural network proxy model is as follows:

[0033] The datasets of primary and secondary inductances are processed and filtered to remove inductance parameters that do not conform to the actual physical laws.

[0034] The dataset is divided into a training dataset and a test dataset, and scaling, data transformation and normalization operations are performed on the training dataset and the test dataset respectively.

[0035] The training dataset is input into the ANN model to train the ANN model, thus obtaining the artificial neural network surrogate model.

[0036] Preferably, the formula for calculating the instantaneous transmission power is:

[0037]

[0038] The optimal shift ratio is obtained by taking the extreme value of D with respect to the instantaneous transmission power formula, that is...

[0039] The phase-shift modulation method for a dual active bridge DC-DC converter provided by this invention has the following beneficial effects:

[0040] By fully considering the nonlinear factors of the high-frequency transformer core in the dual active bridge DC-DC converter, especially since the core of the large-capacity DC-DC converter frequently operates in the nonlinear saturation range, the permeability is dynamically adjusted according to the operating point of the high-frequency transformer to obtain the corresponding dynamically changing inductance. Through the dynamic inductance, the power transmission efficiency of the high-frequency transformer can be obtained, and the optimal phase shift ratio can be obtained, thus formulating a single-phase shift modulation strategy that considers the nonlinear factors of the core. Attached Figure Description

[0041] To more clearly illustrate the embodiments and design schemes of the present invention, the accompanying drawings required for this embodiment will be briefly described below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a flowchart of the phase-shifting modulation method of the dual active bridge DC-DC converter according to an embodiment of the present invention;

[0043] Figure 2 It is a dual active bridge DC-DC converter circuit;

[0044] Figure 3 The basic magnetization curve;

[0045] Figure 4 This is an artificial neural network model with a multilayer perceptron.

[0046] Figure 5 For the training process of a single artificial neuron;

[0047] Figure 6 This describes the training process for artificial neural networks. Detailed Implementation

[0048] Example

[0049] This invention provides a phase-shift modulation method for a dual active bridge DC-DC converter, comprising the following steps: Figure 1 As shown:

[0050] Step 1: Using the Latin hypercube sampling method, the design variables of the high-frequency transformer of the converter, primary voltage V1, secondary voltage V2, and shift ratio D, are randomly sampled in the design variable sample space to obtain multiple different combinations of design variables V1, V2, and D.

[0051] The methods for obtaining multiple different combinations of design variables V1, V2, and D are as follows:

[0052] (1) Sampling is performed in the three-dimensional variable space consisting of the primary voltage V1, secondary voltage V2 and shift ratio of the design variables;

[0053] (2) Divide each dimension into m non-overlapping intervals of equal length;

[0054] (3) Randomly select a point in each interval of each dimension;

[0055] (4) The randomly selected points are combined into a vector to form multiple variable combinations.

[0056] Step 2: Calculate the initial equivalent inductance of the primary and secondary windings based on the initial permeability of the high-frequency transformer and the finite element control equations.

[0057] The initial permeability is the slope of the tangent to the fundamental magnetization curve at the origin, as shown in the fundamental magnetization curve. Figure 2 As shown.

[0058] The governing equations for magnetic field calculation are:

[0059]

[0060]

[0061]

[0062] μ is the permeability, determined by the basic magnetization curve, A is the vector magnetic potential, σ is the conductivity, φ is the scalar potential, and J is the current density. The current density of the dual active bridge DC-DC converter is calculated from the primary and secondary currents at different time steps.

[0063] The permeability μ = f(B) is obtained from a table based on the basic magnetization curve. That is, the relationship between permeability μ and magnetic flux density B is established based on the basic magnetization curve, while the relationship between magnetic flux density and vector magnetic potential is...

[0064] Then, the initial equivalent inductance of the primary and secondary windings is calculated using the energy method. The calculation formula is as follows:

[0065]

[0066] In the formula, W is the magnetic field energy, and I... pt / st Let be the current in the primary and secondary windings at time t. During the initial equivalent inductance calculation at t=0, this current can be applied as a value close to 0.

[0067] The formula for calculating magnetic field energy is: Here, V represents the computational region.

[0068] Step 3: Calculate the initial primary and secondary current values ​​for the initial time step based on the initial equivalent inductance of the primary and secondary windings. Use these primary and secondary current values ​​as the current excitation for the high-frequency transformer to obtain the permeability for the next time step. The dual active bridge DC-DC converter circuit is shown below. Figure 3 As shown.

[0069] The current flowing through the equivalent inductance L (the total inductance of the primary and secondary windings referred to the primary side) is:

[0070]

[0071] In the formula, i L Let i be the current flowing through inductor L. p i s V1 and V2 are the primary and secondary currents, respectively, N is the turns ratio, and V1 and V2 are the primary and secondary voltages, respectively.

[0072] Step 4: Use the field-circuit coupling method to iteratively calculate Step 3 at different time steps (the permeability is no longer the initial permeability) to obtain the equivalent inductance of the primary and secondary windings at different time steps. Determine whether the current time is less than the set simulation time T. If it is less than the value, repeat Step 2 and Step 3 until the current time is greater than or equal to the set simulation time T.

[0073] Step 5: Construct a large dataset of primary and secondary inductance using different combinations of variables and inductance values ​​at each time step.

[0074] Step 6: Based on a large dataset of primary and secondary inductances, construct a variable combination (V) using a multilayer perceptron artificial neural network. 1,n V 2,n D n ,t) with primary inductance and secondary inductance (L p,t L s,t Artificial neural network proxy model.

[0075] like Figure 4 As shown, the training process of a single artificial neuron in a multilayer perceptron artificial neural network, consisting of primary voltage V1, secondary voltage V2, shift ratio D, and time t, is as follows: Figure 5 As shown.

[0076] Wherein, the activation function φ is the Sigmoid function, x1~x4 represent the four inputs (i.e., V1, V1, D, t), w i (i = 1, 2, 3, 4) represents the weighting coefficient, b represents the bias, and Y is the output (i.e., inductance L). Thus, the primary and secondary inductances L can be established. p / s Artificial neural network surrogate model function (V1, V2, D, t).

[0077] In this invention, the training process of the artificial neural network surrogate model is as follows: Figure 6As shown, based on the large dataset of primary and secondary equivalent inductances calculated by the finite element model, data aggregation is performed, and inductance parameters that do not conform to the actual physical laws are removed and filtered out. Then, the large dataset of inductances is divided into training and testing datasets, and scaling, data transformation, and normalization operations are performed on the training and testing datasets respectively. Based on this, the ANN data is trained, i.e., the weight coefficients and biases are optimized. To prevent overfitting, the ANN model needs to be evaluated. After the ANN model evaluation meets the requirements, the scaled data is restored, and the performance of the ANN model is evaluated using a test machine. The scaling factor and ANN data are stored for predicting future inductance values, and the predicted inductance values ​​are used for optimizing the dual active bridge single-phase-shift modulation strategy in the next step.

[0078] Step 7: The inductance value at each time step can be quickly calculated using an artificial neural network surrogate model. The instantaneous transmission power model of the equivalent inductance (the total inductance of the primary and secondary inductances referred to the primary side) can be determined by taking the derivative of the instantaneous transmission power model with respect to D and finding its extreme value. The optimal shift ratio D can then be determined.

[0079] The formula for calculating instantaneous power is:

[0080]

[0081] In the formula, Δt is the step size, j represents time j, and j+1 represents time j+1;

[0082] This expression is a function of the shift ratio D. The optimal shift ratio D can be determined by finding the extrema through partial derivatives, i.e.

[0083] As described above, the inductance value can be quickly obtained through the surrogate model of the artificial neural network, thus obtaining the optimal phase shift ratio. Therefore, this method has good real-time performance and computational accuracy, and can be used for the phase shift modulation strategy optimization of large-capacity DC-DC converters.

[0084] By fully considering the nonlinear factors of the high-frequency transformer core in dual active bridge DC-DC converters, especially since the core of large-capacity DC-DC converters frequently operates in the nonlinear saturation range, the permeability is dynamically adjusted according to the operating point of the high-frequency transformer to obtain the corresponding dynamically changing inductance. This allows for the development of a single-phase-shift modulation strategy that takes into account the core's nonlinearity. This improves the power transmission efficiency and provides a more precise and optimized phase-shift modulation strategy for large-capacity dual active bridge DC-DC converter applications. However, current dual active bridge DC-DC converters treat the high-frequency transformer as a linear inductor in their power electronic switch control strategies, neglecting the nonlinear factors of the core.

[0085] The above-described embodiments are merely preferred embodiments of the present invention, and the scope of protection of the present invention is not limited thereto. Any simple changes or equivalent substitutions of the technical solutions that can be obviously obtained by those skilled in the art within the scope of the technology disclosed in the present invention shall fall within the scope of protection of the present invention.

Claims

1. A phase-shift modulation method for a dual active bridge DC-DC converter, characterized in that, Includes the following steps: Latin hypercube sampling method is used to measure the primary voltage of the high-frequency transformer in a DC-DC converter. V 1. Secondary voltage V 2 compared to moving D Randomly sample within the design variable sample space to obtain V 1. V 2 and D A combination of multiple different variables; The initial equivalent inductance of the primary and secondary windings is calculated based on the initial permeability of the high-frequency transformer and the finite element control equations. The initial time step primary current value and secondary current value are calculated based on the initial equivalent inductance of the primary and secondary windings, and the primary and secondary current values ​​are used as the current excitation of the high-frequency transformer to obtain the permeability of the next time step. The field-circuit coupling method is used to iterate the permeability of the next time step in order to calculate the equivalent inductance of the primary and secondary windings at different time steps, thereby obtaining a dataset of primary and secondary inductances corresponding to multiple different combinations of variables at different time steps. The dataset is input into a multilayer perceptron artificial neural network to construct an artificial neural network surrogate model of the variable combination and the primary and secondary inductances; The artificial neural network proxy model is used to calculate the primary and secondary inductances of the high-frequency transformer at different time steps in real time, and the instantaneous transmission power of the high-frequency transformer based on the equivalent circuit of the dual active bridge DC-DC converter is obtained based on the primary and secondary inductances of the high-frequency transformer at different time steps. The real-time optimal shift ratio of the high-frequency transformer is determined based on the instantaneous transmission power. D opt ; The process of constructing the artificial neural network proxy model is as follows: The datasets of primary and secondary inductances are processed and filtered to remove inductance parameters that do not conform to the actual physical laws. The dataset is divided into a training dataset and a test dataset, and the training dataset and the test dataset are scaled, transformed and normalized respectively. The training dataset is input into a multilayer perceptron artificial neural network to train the multilayer perceptron artificial neural network, thereby obtaining an artificial neural network surrogate model; The formula for calculating the instantaneous transmission power is: In the formula, △ t Step size, j express j Time Step j+1 express j +1 step, N is the turns ratio; The optimal shift ratio D opt Through instantaneous power transmission D The derivative and the extreme value are obtained, that is... .

2. The phase-shifting modulation method for a dual active bridge DC-DC converter according to claim 1, characterized in that, The method for obtaining multiple combinations of the aforementioned variables is as follows: At primary voltage V 1. Secondary voltage V 2 compared to moving D Random sampling within the design variable sample space; Divide each dimension into non-overlapping sections of equal length. m A range; Randomly select a point in each interval of each dimension; The randomly selected points are combined into a vector, forming multiple combinations of variables.

3. The phase-shifting modulation method for a dual active bridge DC-DC converter according to claim 1, characterized in that, The initial permeability is determined based on the basic magnetization curve, and the initial permeability is the slope of the tangent line to the basic magnetization curve at the origin.

4. The phase-shift modulation method for a dual active bridge DC-DC converter according to claim 1, characterized in that, The governing equation is: in, μ Permeability, A For vector magnetic potential, σ For electrical conductivity, ϕ For scalar potential, J denoted as current density.

5. The phase-shift modulation method for a dual active bridge DC-DC converter according to claim 4, characterized in that, The formulas for calculating the initial equivalent inductance of the primary and secondary windings are as follows: In the formula, W For magnetic field energy, I pt / st For the primary and secondary windings in t The current of the time step; The magnetic field energy W The calculation formula is: , In the formula, V Indicates the calculation area.

6. The phase-shift modulation method for a dual active bridge DC-DC converter according to claim 5, characterized in that, The formulas for calculating the primary current value and the secondary current value are as follows: , i p = i L , i s = Ni p In the formula, i L For flow through the inductor L The current, i p , i s These are the primary current value and the secondary current value, respectively. N The turns ratio, V 1 and V 2 represents the primary voltage and the secondary voltage, respectively.

Citation Information

Patent Citations

  • Multi-objective optimization control method for full power range of dual-active bridge converter

    CN113258785A

  • Dual active bridge converter

    CN116032130A