Model predictive control method and device for dual active bridge converter

Abstract

The present invention discloses a model predictive control method and device for a dual active bridge converter, which is used to solve the problems of low transmission efficiency, large power conversion loss and large transformer current stress in existing related methods. It includes: obtaining a predictive control model of a dual active bridge converter, a target cost function and a mobile discrete control set at the end of the dth cycle, the target cost function is constructed based on the minimum voltage prediction error, the minimum voltage change and the minimum current stress as control constraints; the mobile discrete control set contains multiple candidate search points, each candidate search point corresponds to a control variable combination; according to the predictive control model, the output voltage prediction value of each candidate search point at the end of the d+2th cycle is calculated; for each candidate search point, the control variable combination and the output voltage prediction value are substituted into the target cost function for search optimization, and the candidate search point that minimizes the target cost function value is used as the optimal working point.

Description

Technical Field

[0001] The present invention relates to the technical field of power electronic converters, and in particular to a model predictive control method for a dual active bridge converter, a model predictive control device for a dual active bridge converter, an electronic device and a storage medium. Background Art

[0002] DC-DC (Direct Current to Direct Current) converters are widely used in power electronic systems such as offshore wind farms, photovoltaic power plants, electric vehicle charging stations, DC microgrids, and energy storage systems. As an interface circuit, the DC-DC converter can connect high and low voltage DC buses to improve power supply quality and energy efficiency. In addition, the DC-DC converter also plays a role in power flow control, voltage regulation, fault isolation, and other functions.

[0003] In actual situations, the isolated bidirectional DC-DC converter dual active bridge circuit (hereinafter referred to as the dual active bridge converter) can meet almost all the requirements of the medium voltage DC system. Among them, the dual active bridge converter can automatically adjust the bidirectional power flow, easily implement soft switching, and reduce power loss. The dual active bridge converter also has the characteristics of electrical isolation, simple and symmetrical circuit structure, high reliability, and robustness to circuit parameter changes. It is more suitable for scenarios with high requirements for electrical safety such as offshore wind farms and electric vehicle charging stations. Not only that, the dual active bridge converter plays an important role in multiple application scenarios such as flexible DC transmission, integrated power supply systems, electric vehicle V2G charging piles, solid-state transformers, etc. Therefore, the research on the dual active bridge converter circuit is of great significance.

[0004] At present, the SPS (Single-Phase Shift), DPS (Dual-Phase Shift), TPS (Triple-Phase Shift) modulation modes of dual active bridge converters and their optimization methods and results have been studied in detail. Model predictive control algorithms have also been applied to the SPS, DPS and TPS modulation of dual active bridge converters, achieving excellent control performance.

[0005] However, the current SPS modulation of the dual active bridge converter does not match the voltage on the primary and secondary sides with the transformer turns ratio, or has low transmission efficiency at low power. Although the EPS (Extended Phase Shift), DPS, and TPS modulation of the dual active bridge converter can expand the soft switching range, the additional circulating current introduced into the H-bridge will increase the power conversion conduction loss and bring greater transformer current stress. Summary of the invention

[0006] The present invention provides a model predictive control method for a dual active bridge converter, a model predictive control device for a dual active bridge converter, an electronic device and a storage medium, which are used to solve or partially solve the technical problems of low transmission efficiency, large power conversion loss and large transformer current stress existing in the existing dual active bridge converter related modulation methods.

[0007] The present invention provides a model predictive control method for a dual active bridge converter, the method comprising:

[0008] Obtaining a predictive control model, an objective cost function, and a moving discrete control set at the end of the dth cycle of the dual active bridge converter, wherein the objective cost function is constructed based on minimum voltage prediction error, minimum voltage change, and minimum current stress as control constraints; the moving discrete control set includes multiple candidate search points, each of which corresponds to a control variable combination;

[0009] Calculate the output voltage prediction value of each candidate search point at the end of the d+2th cycle according to the prediction control model;

[0010] For each candidate search point, the control variable combination and the output voltage prediction value are substituted into the target cost function for search optimization calculation, and the candidate search point corresponding to the control variable combination that minimizes the target cost function value is used as the optimal operating point at the end of the d+2th cycle.

[0011] The present invention also provides a model predictive control device for a dual active bridge converter, comprising:

[0012] A prediction data acquisition module is used to obtain a prediction control model of a dual active bridge converter, a target cost function and a moving discrete control set at the end of the dth cycle, wherein the target cost function is constructed based on minimum voltage prediction error, minimum voltage change and minimum current stress as control constraints; the moving discrete control set includes a plurality of candidate search points, each of which corresponds to a control variable combination;

[0013] An output voltage prediction value calculation module, used for calculating the output voltage prediction value of each candidate search point at the end of the d+2th cycle according to the prediction control model;

[0014] The search optimization calculation module is used to substitute the control variable combination and the output voltage prediction value into the target cost function for each candidate search point to perform search optimization calculation, and take the candidate search point corresponding to the control variable combination that minimizes the target cost function value as the optimal operating point at the end of the d+2th cycle.

[0015] The present invention also provides an electronic device, the device comprising a processor and a memory:

[0016] The memory is used to store program codes and transmit the program codes to the processor;

[0017] The processor is used to execute the model predictive control method of the dual active bridge converter as described in any one of the above items according to the instructions in the program code.

[0018] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium is used to store program code, and the program code is used to execute the model predictive control method of the dual active bridge converter as described in any one of the above items.

[0019] It can be seen from the above technical solutions that the present invention has the following advantages:

[0020] A model predictive control method for a dual active bridge converter is provided. First, the predictive control model, target cost function and mobile discrete control set at the end of the dth cycle of the dual active bridge converter are obtained. The target cost function is constructed based on the minimum voltage prediction error, the minimum voltage change and the minimum current stress as control constraints; the mobile discrete control set contains multiple candidate search points, each candidate search point corresponds to a control variable combination; then, according to the predictive control model, the output voltage prediction value of each candidate search point at the end of the d+2th cycle is calculated; for each candidate search point, the control variable combination and the output voltage prediction value are substituted into the target cost function for search optimization, and the candidate search point that minimizes the target cost function value is taken as the optimal working point. Therefore, by using the target cost function with multiple control constraints and combining the set mobile discrete control set to search and optimize the optimal solution, the optimal working point can be quickly determined, accurate and fast voltage tracking and minimization of current stress can be achieved, and power conversion loss can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0022] Figure 1 It is a circuit principle schematic diagram of a dual active bridge converter;

[0023] Figure 2 A flow chart of steps of a model predictive control method for a dual active bridge converter;

[0024] Figure 3 A schematic diagram of a reduced-order model of a dual active bridge converter;

[0025] Figure 4 is a power model p=F1(D,f s ) three-dimensional diagram and contour line (top view);

[0026] Figure 5 It is a three-dimensional diagram and contour line (top view) schematic diagram of a current stress model;

[0027] Figure 6 It is a curve diagram showing the minimum current stress under a given power;

[0028] Figure 7 It is a schematic diagram when the current peak contour line is tangent to the parabola at the outward shift phase angle;

[0029] Figure 8 It is a schematic diagram of the change of transmission power-external shift phase angle under different dead time;

[0030] Fig. 9 It is a schematic diagram of the change of the current peak value with the external shift phase angle under a given power;

[0031] Fig.10 It is a schematic diagram of the change of current peak value with switching frequency under a given power;

[0032] Fig.11 A schematic diagram of a model predictive control process for a dual active bridge converter;

[0033] Fig.12 A schematic diagram of a model prediction optimization process for a dual active bridge converter;

[0034] Fig.13 A schematic diagram of the overall flow of a model predictive control method for a dual active bridge converter;

[0035] Fig.14 The structure block diagram of a model predictive control device for a dual active bridge converter. DETAILED DESCRIPTION

[0036] The embodiments of the present invention provide a model predictive control method for a dual active bridge converter, a model predictive control device for a dual active bridge converter, an electronic device and a storage medium, which are used to solve or partially solve the technical problems of low transmission efficiency, large power conversion loss and large transformer current stress existing in the existing dual active bridge converter related modulation methods.

[0037] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] As an example, at present, the SPS, DPS, TPS modulation modes of dual active bridge converters and their optimization methods and results have been studied in detail. Model predictive control algorithms have also been applied to the SPS, DPS and TPS modulation of dual active bridge converters, achieving excellent control performance.

[0039] However, the current SPS modulation of the dual active bridge converter does not match the voltage on the primary and secondary sides with the transformer turns ratio, or the transmission efficiency is low at low power. Although the EPS, DPS, and TPS modulation of the dual active bridge converter can expand the soft switching range, the additional circulating current introduced into the H-bridge will increase the power conversion conduction loss and bring greater transformer current stress.

[0040] Based on the above problems, the present invention further studies and analyzes and believes that if an adaptive step size can be introduced into the model predictive control, and variable frequency variable step size modulation is performed in the process of solving the optimal solution of the working point based on the adaptive step size, the system can have fast transition performance and response speed. The variable frequency modulation of the dual active bridge converter can be combined with phase shift modulation without changing the working mode to expand the soft switching range to a lower power level. In addition, variable frequency modulation can increase the output power range of the dual active bridge converter, and can also make full use of the advantages of low current stress at high frequency and high output power at low frequency.

[0041] Therefore, one of the core invention points of the embodiment of the present invention is: in view of the deficiencies in the prior art, a variable frequency and variable step size model predictive control method for a dual active bridge converter is provided. First, in order to improve the power conversion efficiency of the dual active bridge converter at low power output and reduce the current stress of the isolation transformer, the embodiment of the present invention uses a method combining numbers and shapes to derive the conditions for minimizing the transformer current stress, and converts the condition into one of the cost functions in the model predictive control algorithm, and designs a unique target cost function in combination with other cost functions to achieve accurate and fast voltage tracking and minimize current stress. Secondly, an adaptive mechanism is added to the optimization process of the discretized model predictive control to dynamically change the control step size of the two variables, the external phase shift angle and the switching frequency, to speed up the transient dynamic performance and improve the steady state stability accuracy. At the same time, considering the influence of the switch dead zone effect and the actual factors such as the transformer core saturation, the upper and lower limits of the two control variables, the external phase shift angle and the switching frequency, are set to reduce the transformer stress at low power output of the dual active bridge converter, reduce the power conversion loss, and achieve fast voltage tracking or output control at the same time, and strike a balance between steady state and dynamic performance. Therefore, through the technical solution provided by the present invention, a variable frequency and variable step size model predictive control strategy for the dual active bridge converter can be formed to achieve accurate and fast voltage tracking, while improving the transient response capability of the circuit and improving the steady-state control accuracy.

[0042] To better explain, Figure 1 The schematic diagram of the circuit principle of the dual active bridge converter used in this embodiment is shown.

[0043] Combination Figure 1 The dual active bridge converter consists of two H-bridges on the input and output sides (composed of four MOSFETs (Metal-Oxide-Semiconductor Field-Effect Transistors) and their parasitic anti-parallel diodes), transformers, and inductors. The leakage inductance of the transformer itself can be used as an inductor. The transformer plays an important role in providing electrical isolation and voltage matching between the primary and secondary sides of the circuit.

[0044] According to the number of control variables and the different operation modes, the modulation of the dual active bridge converter can be divided into many different modes. In the general modulation mode of the dual active bridge converter, there are generally four control variables: the outer phase shift angle between the two H bridges on both sides, the inner phase shift angle between the two bridge arms inside the primary H bridge, the inner phase shift angle between the two bridge arms inside the secondary H bridge, and the switching frequency.

[0045] In order to achieve stable output and fast response of dual active bridge converter, a suitable control method needs to be designed. At present, the most common control method is PI control with only feedback. This method has low computational complexity and low implementation cost. However, the dynamic performance is weak, and it is easy to show performance degradation when the operating point changes. Output current feedforward control (OCFF) can significantly reduce the output impedance and has good load current interference suppression capability. However, output current feedforward control is very dependent on the selection of circuit parameters, and the actual performance is often not as good as the expected design. Disturbance-Observer-Based Control (DOBC) based on impedance observer first uses the observer to estimate the total disturbance and uncertainty, and then uses the estimation results to generate corresponding compensation, so it can achieve excellent control performance. However, the control method based on impedance observer is not sensitive to changes in circuit parameters and is prone to control errors. Feedforward current control (FFCC) can control the peak current and has good load interference suppression capability, but it is sensitive to noise. Predictive current control has strong dynamic performance, but each switching cycle requires AC sampling and calculation, which may not be suitable for high-frequency applications.

[0046] In recent years, Model Predictive Control (MPC) has been often considered for use in the control of power electronic converters due to its advantages such as easy inclusion of nonlinearities and constraints, fast dynamic speed, and simple digital implementation. In general, the design and control of the model predictive control algorithm is relatively simple, with excellent dynamic performance and strong anti-interference ability. The use of a microprocessor with strong computing power can make up for the disadvantage of high computational complexity, making it very suitable for use in dual active bridge converters.

[0047] Based on model predictive control, Figure 2 , shows a flow chart of the steps of a model predictive control method for a dual active bridge converter provided by an embodiment of the present invention, which may specifically include the following steps:

[0048] Step 201, obtaining a predictive control model, a target cost function and a moving discrete control set at the end of the dth cycle of the dual active bridge converter, wherein the target cost function is constructed based on minimum voltage prediction error, minimum voltage change and minimum current stress as control constraints; the moving discrete control set includes multiple candidate search points, each of which corresponds to a control variable combination;

[0049] In practical applications, before obtaining relevant models or data for predictive control, it is necessary to first build a predictive control model, design a cost function, and generate relevant data sets for prediction.

[0050] For the predictive control model, various methods have been proposed in the relevant field for modeling the dual active bridge converter, and various models with different accuracy and complexity have been obtained. Among them, the reduced-order model has a good trade-off between accuracy and complexity. Therefore, in the embodiment of the present invention, a predictive model based on the reduced-order model of the dual active bridge converter is used as the predictive control model.

[0051] Assume that the current moment is the end moment of the dth switching cycle, and the output voltage measurement value V2[d] at the current moment is obtained through measurement, and the control variable combination of the upcoming d+1th cycle, that is, the external phase angle and the switching frequency (D[d+1], f s [d+1]) is given at the end of the d-1th cycle.

[0052] First, the dynamic equation of the capacitor needs to be obtained based on the reduced-order model of the dual active bridge converter circuit; then the dynamic equation of the capacitor is discretized to obtain the output voltage V2[d+1] at the end of the d+1th cycle; then the predictive control model of the model predictive control algorithm is obtained from the output current of the three adjacent cycles.

[0053] Specifically, the process of constructing the predictive control model can be implemented by executing the following sub-steps S01 to S03:

[0054] Step S01: constructing a circuit reduction model of a dual active bridge converter, and deriving a capacitor dynamic equation based on the circuit reduction model;

[0055] Exemplarily, the circuit schematic diagram of the reduced-order model of the dual active bridge converter is as follows: Figure 3 The relationship given by the reduced-order model is shown in equation (1):

[0056]

[0057] Among them, L and V1 represent side 1 (i.e. Figure 3 (center left) Circuit inductance and DC source voltage; n represents transformer ratio; D represents external shift phase angle; f s Indicates the switching frequency.

[0058] Based on the above formula (1), the dynamic equation of capacitor C2 can be obtained as shown in the following formula (2):

[0059]

[0060] Where, t represents time; V2 represents output voltage; T s Represents a time period.

[0061] Step S02: discretize the capacitor dynamic equation to obtain the current output voltage prediction value at the end of the d+1th cycle;

[0062] Based on the average value of a switching cycle, the predictive control algorithm will only make an operation and decision once at the boundary of the switching cycle. Therefore, by discretizing the above formula (2), we can get the expression of the current output voltage prediction value at the end of the d+1th cycle as shown in the following formula (3):

[0063]

[0064] Wherein, V2[d] represents the output voltage at the end of the d-th switching cycle; similarly, D[d] represents the shift phase of the d-th switching cycle, and D[d] is determined by the optimization algorithm at the end of the d-1-th switching cycle.

[0065] Step S03: Assuming that the output currents at the end of the dth cycle, the end of the d+1th cycle, and the end of the d+2th cycle are the same, derive the predictive control model corresponding to the output voltage of the dual active bridge converter at the end of the d+2th cycle based on the current output voltage prediction value.

[0066] Due to the symmetry between the first half cycle and the second half cycle, the transmission power of the first and second half cycles is equal. Therefore, the average transmission power (forward transmission) of the dual active bridge converter can be calculated through the half cycle, and the expression is shown in the following formula (4):

[0067]

[0068] Combining formula (1), the expression of transmission power P can be obtained as shown in formula (5):

[0069]

[0070] The reference value P shown in the following formula (6) N :

[0071]

[0072] By normalizing equation (5), the normalized value p of the forward transmission power can be obtained as shown in equation (7):

[0073]

[0074] In order to leave enough computing time for the DSP module, it is necessary to complete the calculation in the DSP at the end of the dth switching cycle and determine the control variable D[d+1], f s[d+1]. The prediction horizon of the model predictive control algorithm is 1, so it is necessary to predict the output voltage value V2[d+2] of the next two cycles. Assume that the output current of the three adjacent cycles does not change much (for ease of calculation, it is assumed that the output current at the end of the dth cycle, the end of the d+1th cycle, and the end of the d+2th cycle are the same):

[0075] I2[d]=I2[d+1]=I2[d+2] (8)

[0076] Then the predictive control model of the model predictive control algorithm is shown in the following formula (9):

[0077]

[0078] Where C2 is the output capacitor; i h2 [d+2] and i h2 [d+1] is the end time of the d+2th cycle and the end time of the d+1th cycle respectively. Figure 3 The current is input on the middle 2 side (i.e. the right side).

[0079] The advantage of model predictive control is that it is easy to include multiple optimization objectives and constraints. Therefore, after the predictive control model is determined, the objective cost function needs to be designed.

[0080] Specifically, the process of constructing the target cost function can be implemented by executing the following sub-steps S11 to S14:

[0081] Step S11: constructing a first cost function with minimum voltage prediction error as a control constraint;

[0082] The first cost function may also be referred to as a basic cost function, which aims to minimize the distance between the output voltage prediction value and the target reference value, so as to make the output voltage approach the target value.

[0083] The first cost function is as follows:

[0084] G1=(V 2,ref -V2[d+2]) 2 (10)

[0085] Where G1 represents the first cost function; V 2,ref represents the target reference value of the output voltage; V2[d+2] represents the predicted value of the output voltage at the end of the d+2th cycle.

[0086] Step S12: constructing a second cost function with minimum voltage change as a control constraint;

[0087] In addition to the basic cost function, a cost function that makes the voltage change small can be reasonably added. The significance of this cost function is to minimize the change between the voltage at the end of the d+2th cycle and the voltage at the end of the dth cycle, tending to make the output voltage change slowly.

[0088] The second cost function is as follows:

[0089] G2=(V2[d]-V2[d+2]) 2 (11)

[0090] Wherein, G2 represents the second cost function; V2[d] represents the output voltage value at the end of the dth cycle.

[0091] Step S13: constructing a third cost function with minimum current stress as a control constraint;

[0092] In addition to the above cost function, a cost function G3 is also designed in the embodiment of the present invention to minimize the current stress of the dual active bridge converter circuit.

[0093] The design process of G3 is as follows: Transmission power model p = F1 (D, f s ) and current stress model i peak,1 =F2(D,f s ) are given in Table 1 below, and their three-dimensional images and contour maps are shown in Table 1. Figure 4 , Figure 5 shown.

[0094]

[0095] Table 1: Calculation expressions for the transmission power model and the current stress model

[0096] When the transmission power p0 is given, the switching frequency expression 1 shown in the following equation (12) can be obtained from the calculation formula of the normalized value of the forward transmission power:

[0097]

[0098] from Figure 4 The power contour diagram shows that at f s -D two-dimensional coordinates, the prediction model is a parabola with the axis of symmetry at D = 0.25. In other words, all combinations (D, f s ) are all on this parabola. Under a given current stress i peak When , the switching frequency expression 2 shown in the following equation (13) can be obtained from equation (8):

[0099]

[0100] from Figure 5The current stress contour diagram shows that at f s -D two-dimensional coordinates, the switching frequency expression 2 is expressed as a linear straight line, whose slope and intercept are both i peak,1 In other words, all the functions that make the current peak i peak,1 Equal to (D,f s ) are all on this straight line, and the position of this straight line is affected by i peak,1 impact.

[0101] like Figure 6 As shown, when the power p0 is given, there are countless (D, f s ) point, each (D,f s ) points correspond to a unique i peak,1 Value, each i peak,1 The value corresponds to a unique straight line. Figure 6 It can be seen that if and only if the straight line of switching frequency expression 2 is tangent to the parabola of switching frequency expression 1, i peak,1 The value of (D,f s ) combination is the combination that minimizes the current stress at a given power p0. That is to say, at a given power p0, when the current stress contour is tangent to the curve, the current stress is minimum at this power p0.

[0102] So the above problem is transformed into an analytical geometry problem: let the quadratic discriminant of the analytical expression of the intersection of the straight line and the parabola be equal to zero, and then we can solve that when the straight line and the parabola are tangent, D, f s ,i peak,1 Constraints between. Combining the switching frequency expression 1 of equation (12) and the switching frequency expression 2 of equation (13), we can get:

[0103]

[0104] Let the discriminant of the quadratic equation of (14) be equal to 0:

[0105]

[0106] The solution is that under the given power p0 and k = V1 / nV2, i peak,1 The minimum values ​​of are as follows:

[0107]

[0108] Formula (16) is the theoretical minimum current peak value under given power p0 and voltage conversion ratio (k ≥ 1). This formula reveals that the more the voltage conversion ratio k deviates from 1, the greater the theoretical minimum current peak value. Substituting formula (16) into formula (14), we get peak,1 The minimum external phase shift angle D must satisfy the following constraints:

[0109]

[0110] From formula (17), we can see that i peak,1 The value of the minimum external phase shift angle D has nothing to do with the given power p0, but only with the current voltage conversion ratio k = V1 / nV2. Substituting equation (17) into equation (12) gives the switching frequency expression

[0111] In formula 1, we can get i peak,1 Minimum switching frequency f s The value of is shown in formula (18):

[0112]

[0113] According to the above results, the third cost function to minimize the current stress can be formulated as follows:

[0114]

[0115] Among them, G3 represents the third cost function; D and f s They represent the control variable combination corresponding to the current stress (D, f s ) in the external phase angle and switching frequency; D * and They represent minimizing the current peak value, that is, approaching the theoretical minimum value of current stress i peak,1,min The external shift phase angle is related to the switching frequency.

[0116] The goal of the third cost function is to make (D,f s ) approaches So that i peak,1 Approaching the theoretical minimum value of current stress i peak,1,min It should be noted that the third cost function can also be expressed in other similar or principle equivalent forms, and is not limited to formula (19).

[0117] It is worth noting that in formula (17), when k→1, D * →0, thus The curve changes as Figure 7 As shown in the figure, the current peak contour line is tangent to the parabola at D = 0, so D * = 0. However, in reality, if forward power is to be transmitted, whether D→0 or f s →0 means that power cannot be transmitted normally.

[0118] For the external phase shift angle D, the actual circuit needs to consider the dead zone effect, that is, the existence of the switching dead zone time will weaken the output power at the external phase shift angle D, which is specifically manifested as follows: Figure 8As shown in the figure, M represents the proportion of dead time in the half cycle.

[0119] In actual situations, due to the existence of parasitic capacitance of the switching device, the dead time of the switch is necessary, otherwise the switching device will be damaged by the large voltage penetration, thus threatening the life of the system. Therefore, if the optimal external phase angle D given by equation (17) is * →0, the forward power cannot be transmitted. Therefore, the minimum value of D must be limited to D min , to transmit forward power normally. D min The specific value of can be determined by theoretical deduction or trial and error.

[0120] Depend on Figure 7 It can be seen that on the parabola of a given power p0, as D increases, the corresponding current peak also increases. Therefore, in the case of k→1, the current peak is roughly positively correlated with D. Therefore, when the minimum value of D is limited, D min After that, the outward phase angle that makes the current peak value take the minimum value is exactly D min .

[0121] For the switching frequency f s , f s Too low will lead to excessive excitation current, saturation of transformer core, and hinder the normal operation of magnetic components. Therefore, it is necessary to limit f s The minimum value of f s，min For specific values, please refer to the transformer data sheet. s，min It is also affected by the rated output power. At the same time, due to the limitations of chip computing power and switching losses, the switching frequency cannot be set too high, so it is necessary to limit f s The maximum value of f s，max .

[0122] Using the same method, we can get the cost function that minimizes the current stress when k < 1. Under a given power p0 and k = V1 / nV2, i peak，2 The minimum value of is shown in the following formula (20):

[0123]

[0124] At this time, i peak，2 The minimum external phase shift angle D must satisfy the following constraints:

[0125]

[0126] Make i peak，2 Minimum switching frequency f s The value of is the same as that of equation (18). Equation (20) is the theoretical minimum current peak value under a given power and voltage conversion ratio (k < 1).

[0127] The complete optimization scheme for minimizing current stress is summarized in Table 2 below:

[0128]

[0129] Table 2: Optimization scheme of the third cost function when minimizing current stress

[0130] It can be seen that replacing k in the expression of the optimal external shift phase angle when k ≥ 1 is The expression can be transformed into the condition of k<1.

[0131] When the power p0 is given, the following formula (22) can be obtained from formula (7):

[0132]

[0133] Formula (22) is D,f s The constraints satisfied between them under a given power p0 can be solved by inversely to obtain the following equation (23):

[0134]

[0135] Since it is obtained that under a given power p0, the current peak i peak,1 The change of the external phase angle D is shown in the following formula (24):

[0136]

[0137] The image of formula (24) is as follows Fig. 9 As shown. There is a unique minimum point in the domain of D. In order to find the minimum point, we only need to peak,1 Find the derivative of D and set it equal to zero:

[0138]

[0139] The minimum point is solved as shown in equation (26):

[0140]

[0141] The above formula (26) is consistent with the above result formula (17), which shows that the two methods are homogeneous. However, the former method draws a three-dimensional function graph, which is more intuitive and clear. Substituting formula (26) into formula (22), we can solve f s，min .

[0142] Similarly, the current peak value i can be plotted under a given power p0. peak，1 With the switching frequency f s The changing image, such as Fig.10 The same method is used to solve f s，min , and then solve D min.

[0143] Step S14: constructing a target cost function according to the first cost function, the second cost function and the third cost function.

[0144] After completing the design of the third cost function G3, since multiple cost functions need their own weight factors to coordinate their influence, the final target cost function can be obtained as follows:

[0145] ct=α1G1+α2G2+α3G3 (27)

[0146] Among them, ct represents the target cost function; α1, α2, and α3 are all weight factors.

[0147] Adjusting the weight factors of each cost function is a tedious task, and the main method currently used is trial and error.

[0148] In an embodiment of the present invention, the process of constructing a mobile discrete control set is as follows: taking the optimal control variable combination at the end of the d+1th cycle predicted at the end of the d-1th cycle as the center, multiple candidate search points are diffused around to form a mobile discrete control set.

[0149] Further, in combination with the foregoing, the optimal control variable combination referred to in the embodiment of the present invention may include the optimal external shift phase angle and the optimal switching frequency at the end of the d+1th cycle predicted at the end of the d-1th cycle. The specific process of forming a mobile discrete control set by spreading multiple candidate search points around the optimal control variable combination at the end of the d+1th cycle predicted at the end of the d-1th cycle can be implemented by executing the following steps S21 to S24:

[0150] Step S21: obtaining a target reference value of the output voltage and the output voltage value at the end of the dth cycle, and calculating an adaptive step length of the diffusion search range according to the target reference value of the output voltage and the output voltage value at the end of the dth cycle;

[0151] Furthermore, the calculation formula of the adaptive step size is:

[0152]

[0153] Among them, V Δ Represents the adaptive step size, corresponding to the voltage difference between the reference voltage value and the actual output voltage value; V 2，ref represents the target reference value of the output voltage; V2[d] represents the output voltage value at the end of the dth cycle; V m Indicates the preset maximum step size, which corresponds to the maximum voltage difference between the reference voltage value and the actual output voltage value.

[0154] Step S22: calculating the adaptive step length of the external phase angle according to the adaptive step length, the first preset basic step length and the first preset adjustment factor, and moving and discretizing the optimal external phase angle around the external phase angle based on the adaptive step length to obtain multiple candidate displacement phase angles;

[0155] Furthermore, the calculation formula of the adaptive step length of the external phase angle is:

[0156] Δ adp =Δ f1 (1+λ1V Δ ) (29)

[0157] Δ adp Represents the adaptive step size of the external phase angle; Δ f1 represents the first preset basic step size; λ1 represents the first preset adjustment factor.

[0158] Step S23: calculating the switching frequency adaptive step length according to the adaptive step length, the second preset basic step length and the second preset adjustment factor, and moving and discretizing the switching frequency adaptive step length around the optimal switching frequency to obtain a plurality of candidate switching frequencies;

[0159] Furthermore, the calculation formula of the switching frequency adaptive step size is:

[0160] Δ adf =Δ f2 (1+λ2V Δ ) (30)

[0161] Δ adf Indicates the switching frequency adaptive step size; Δ f2 represents the second preset basic step size; λ2 represents the second preset adjustment factor.

[0162] Step S24: Based on the multiple candidate displacement phase angles and the multiple candidate switching frequencies, multiple candidate search points are determined to form a moving discrete control set corresponding to the end time of the dth period.

[0163] The adjustment factor (the first preset adjustment factor corresponding to the displacement phase angle and the second preset adjustment factor corresponding to the switching frequency) determines the voltage difference V Δ The influence on the step length, that is, the fitness.

[0164] For the adaptive step size Δ of the external phase angle adp , since the value range of the external phase shift angle D3 when transmitting forward power is (0,0.5), the search range can be linearly expanded. For example, the candidate control set corresponding to the external phase shift angle can be set to (D3,D3+Δ adp ,D3+2Δ adp…), corresponding to multiple candidate displacement phase angles D3, D3+Δ adp ,D3+2Δ adp ….

[0165] For the adaptive step size Δ of the switching frequency adf , due to the switching frequency f s The range of values ​​is large, and the larger the frequency value, the more suitable it is to span a larger step size than the smaller frequency value. Therefore, the switching frequency f s The search range is suitable for exponential expansion. For example, the candidate control set corresponding to the switching frequency can be set to Corresponding to multiple candidate switching frequencies

[0166] Through the above-mentioned adaptive variable frequency variable step size adjustment method, combined with the target cost function for search optimization calculation, when the current voltage deviates far from the target voltage, the search step size can be increased to expand the search range and improve the system's mobility and rapid response capabilities. At the same time, when the current voltage is close to the target voltage, the search step size can be reduced to perform a more refined search and accelerate the convergence speed of the system.

[0167] Among them, the candidate displacement phase angle D3,D3+Δ adp ,D3+2Δ adp …with candidate switching frequencies There is not a strict one-to-one correspondence between them, that is, D3 does not correspond to f s , D3+Δ adp correspond This relationship is a random combination relationship as shown in Table 3 below:

[0168]

[0169] Table 3: Moving discrete control set when candidate search point μ = 3 and random search point r = 3

[0170] Step 202: Calculate the output voltage prediction value of each candidate search point at the end of the d+2th cycle according to the prediction control model.

[0171] The purpose of steps 202 to 203 is to predict the output voltage value V2[d+2] at the end of the d+2th cycle according to the predictive control model and the moving discrete control set, and to determine the optimal operating point.

[0172] In this step, the output voltage prediction value V2[d+2] of each candidate search point in the moving discrete control set at the end of the d+2th cycle can be calculated according to the predictive control model shown in formula (9).

[0173] Step 203, for each of the candidate search points, substitute the control variable combination and the output voltage prediction value into the target cost function for search optimization calculation, and use the candidate search point corresponding to the control variable combination that minimizes the target cost function value as the optimal operating point at the end of the d+2th cycle.

[0174] In this step, the optimal solution of the model predictive control process can be found through search optimization.

[0175] First, the model predictive control process of the dual active bridge converter is as follows Fig.11 Assume that the current moment is the end of the dth switching cycle, the output voltage measurement value V2[d] at the current moment is obtained through measurement, and the control variable combination (D[d+1], f s [d+1]) has been given at the end of the d-1th cycle. The current task is to find the best control variable combination (D[d+2], f[d+1]) that minimizes the target cost function based on the known values ​​and the set mobile discrete control set. s [d+2]), and (D[d+2],f s The search point corresponding to [d+2]) is taken as the optimal working point.

[0176] The optimization process of model predictive control is as follows: Fig.12 As shown in Figure 2, the optimization process of model predictive control is actually a search optimization calculation process based on multiple candidate search points. Combined with the previous content, it can be seen that the control variable combination (D[d+1], f s [d+1]) is already determined. In order to determine the optimal control variable combination (D[d+2], f s [d+2]), in the process of solving the optimal solution of the target cost function, we can refer to the construction process of the mobile discrete control set in the previous step to (D[d+1],f s [d+1]) is the center and spreads μ candidate search points around (i.e. Fig.12 The control variable combinations corresponding to these candidate search points are substituted into the target cost function to find the candidate search point that minimizes the target cost function as the optimal working point.

[0177] In addition, as the complexity of the target cost function increases, the function graph may have multiple local optimal solutions. In order to increase the probability of jumping out of the local optimal solution and finding the global optimal solution, the embodiment of the present invention adds r random search points (i.e. Fig.12 The purple cross in ( ) is used as a candidate random control variable combination. Based on the candidate search points, a new random search point is added. Taking μ=3, r=3 as an example, the mobile discrete control set is shown in the above Table 3.

[0178] In a specific implementation, multiple random search points can be added in addition to multiple candidate search points during the search optimization calculation process; multiple random search points and multiple candidate search points can be combined to perform search optimization and solution to increase the probability of escaping from a local optimal solution and determining a global optimal solution.

[0179] It should be noted that, in order to enable those skilled in the art to better distinguish data of the same type but with different actual meanings, some technical features are distinguished and described using terms such as first and second in the embodiments of the present invention. Terms such as first and second are only used to distinguish data and have no other special meanings. It can be understood that the present invention is not limited to this.

[0180] In an embodiment of the present invention, a variable frequency and variable step size model predictive control method for a dual active bridge converter is provided. First, in order to improve the power conversion efficiency of the dual active bridge converter at low power output and reduce the current stress of the isolation transformer, the embodiment of the present invention uses a method combining numbers and shapes to derive the conditions for minimizing the transformer current stress, and converts the condition into one of the cost functions in the model predictive control algorithm, and designs a unique target cost function in combination with other cost functions to achieve accurate and fast voltage tracking and minimize current stress. Secondly, an adaptive mechanism is added to the optimization process of the discretized model predictive control to dynamically change the control step size of the two variables, the external phase shift angle and the switching frequency, to speed up the transient dynamic performance and improve the steady state stability accuracy. At the same time, considering the influence of the switch dead zone effect and the actual factors such as the transformer core saturation, the upper and lower limits of the two control variables, the external phase shift angle and the switching frequency, are set to reduce the transformer stress at low power output of the dual active bridge converter, reduce the power conversion loss, and achieve fast voltage tracking or output control at the same time, so as to achieve a balance between steady state and dynamic performance. Therefore, through the technical solution provided by the embodiment of the present invention, a variable frequency and variable step size model predictive control strategy for the dual active bridge converter can be formed to achieve accurate and fast voltage tracking, while improving the transient response capability of the circuit and improving the steady-state control accuracy.

[0181] For better explanation, refer to Fig.13 , showing an overall flow diagram of a model predictive control method for a dual active bridge converter provided by an embodiment of the present invention. It should be pointed out that this embodiment only briefly describes the general process of the model predictive control of the dual active bridge converter, and the specific implementation process of each step can be understood by referring to the relevant content in the aforementioned embodiment, which will not be described here. It can be understood that the present invention is not limited to this.

[0182] Construct a predictive control model for dual active bridge converter;

[0183] Taking the minimum voltage prediction error, the minimum voltage variation and the minimum current stress as control constraints, the objective cost function is designed;

[0184] According to the voltage difference between the output voltage value and the target voltage value at the end of the dth cycle, the external phase shift angle D and the switching frequency f are determined respectively. s The adaptive step size of , the larger the gap, the larger the step size;

[0185] Combined with the adaptive step size, the control variable combination (D[d+1], f s [d+1]) is used as the center for two-dimensional expansion, multiple candidate search points are spread out to the surroundings, and the mobile discrete control set at the current moment is constructed;

[0186] Traverse the mobile discrete control set and calculate the output voltage prediction value V of each candidate search point at the end of the d+2th cycle according to the predictive control model. 2_p , and combined with the error compensation algorithm to predict the output voltage V 2_p make corrections;

[0187] For each candidate search point, the control variable combination and the corrected output voltage prediction value V 2_p Substitute it into the target cost function for search optimization calculation, and use the candidate search point corresponding to the control variable combination that minimizes the target cost function value as the optimal working point at the end of the d+2th cycle.

[0188] Reference Fig.14 , shows a structural block diagram of a model predictive control device for a dual active bridge converter provided by an embodiment of the present invention, which may specifically include:

[0189] The prediction data acquisition module 1401 is used to obtain the prediction control model of the dual active bridge converter, the target cost function and the moving discrete control set at the end of the dth cycle, wherein the target cost function is constructed based on the minimum voltage prediction error, the minimum voltage change and the minimum current stress as control constraints; the moving discrete control set includes multiple candidate search points, each of which corresponds to a control variable combination;

[0190] An output voltage prediction value calculation module 1402 is used to calculate the output voltage prediction value of each candidate search point at the end of the d+2th cycle according to the prediction control model;

[0191] The search optimization calculation module 1403 is used to substitute the control variable combination and the output voltage prediction value into the target cost function for each candidate search point to perform search optimization calculation, and take the candidate search point corresponding to the control variable combination that minimizes the target cost function value as the optimal operating point at the end of the d+2th cycle.

[0192] In an optional embodiment, the device further includes:

[0193] A capacitor dynamic equation derivation module is used to construct a circuit reduction model of a dual active bridge converter and derive a capacitor dynamic equation based on the circuit reduction model;

[0194] A capacitor dynamic equation discretization module is used to discretize the capacitor dynamic equation to obtain a current output voltage prediction value at the end of the d+1th cycle;

[0195] The prediction control model derivation module is used to derive the prediction control model corresponding to the output voltage of the dual active bridge converter at the end of d+2 cycles according to the current output voltage prediction value.

[0196] In an optional embodiment, the device further includes:

[0197] The first cost function construction module is used to construct a first cost function with the minimum voltage prediction error as a control constraint. The first cost function is as follows:

[0198]

[0199] Where G1 represents the first cost function; V 2,ref represents the target reference value of the output voltage; V2[d+2] represents the predicted value of the output voltage at the end of the d+2th cycle;

[0200] The second cost function construction module is used to construct a second cost function with minimum voltage change as a control constraint. The second cost function is as follows:

[0201] G2=(V2[d]-V2[d+2]) 2

[0202] Wherein, G2 represents the second cost function; V2[d] represents the output voltage value at the end of the dth cycle;

[0203] The third cost function construction module is used to construct a third cost function with the minimum current stress as the control constraint. The third cost function is as follows:

[0204]

[0205] Among them, G3 represents the third cost function; D and f s They represent the control variable combination corresponding to the current stress (D, f s ) in the external phase shift angle and switching frequency; D * and They represent minimizing the current peak value, that is, approaching the theoretical minimum value of current stress i peak,1,min The external phase shift angle and switching frequency;

[0206] The target cost function construction module is used to construct a target cost function according to the first cost function, the second cost function and the third cost function. The target cost function is as follows:

[0207] ct=α1G1+α2G2+α3G3

[0208] Among them, ct represents the target cost function; α1, α2, and α3 are all weight factors.

[0209] In an optional embodiment, the device further includes:

[0210] The moving discrete control set construction module is used to diffuse multiple candidate search points around the optimal control variable combination at the end of the d+1th period predicted at the end of the d-1th period to form a moving discrete control set.

[0211] In an optional embodiment, the optimal control variable combination includes the predicted optimal external phase shift angle and optimal switching frequency at the end of the d+1th cycle, and the mobile discrete control set building module includes:

[0212] An adaptive step length calculation module, used to obtain a target reference value of the output voltage and an output voltage value at the end of the d-th cycle, and calculate an adaptive step length of the diffusion search range according to the target reference value of the output voltage and the output voltage value at the end of the d-th cycle;

[0213] A candidate shift phase angle generation module is used to calculate the adaptive step length of the external phase angle according to the adaptive step length, the first preset basic step length and the first preset adjustment factor, and to obtain a plurality of candidate shift phase angles by moving and discretizing around the optimal external phase angle based on the adaptive step length of the external phase angle with the optimal external phase angle as the center;

[0214] a candidate switching frequency generating module, configured to calculate a switching frequency adaptive step length according to the adaptive step length, a second preset basic step length and a second preset adjustment factor, and to obtain a plurality of candidate switching frequencies by moving and discretizing the switching frequency adaptive step length around the optimal switching frequency;

[0215] The candidate search point determination module is used to determine multiple candidate search points based on the multiple candidate displacement phase angles and the multiple candidate switching frequencies to form a moving discrete control set corresponding to the end time of the dth period.

[0216] In an optional embodiment, the calculation formula of the adaptive step size is:

[0217]

[0218] The calculation formula of the external phase angle adaptive step size is:

[0219] Δ adp =Δ f1 (1+λ1V Δ )

[0220] The calculation formula of the switching frequency adaptive step size is:

[0221] Δ adf =Δ f2 (1+λ2V Δ )

[0222] Among them, V Δ Represents the adaptive step size, corresponding to the voltage difference between the reference voltage value and the actual output voltage value; V 2,ref represents the target reference value of the output voltage; V2[d] represents the output voltage value at the end of the dth cycle; V m Indicates the preset maximum step size, corresponding to the maximum voltage difference between the reference voltage value and the actual output voltage value; Δ adp Represents the adaptive step size of the external phase angle; Δ f1 represents the first preset basic step length; λ1 represents the first preset adjustment factor; Δ adf Indicates the switching frequency adaptive step size; Δ f2 represents the second preset basic step size; λ2 represents the second preset adjustment factor.

[0223] In an optional embodiment, the device further includes:

[0224] A random search point adding module, used to add a plurality of random search points on the basis of the plurality of candidate search points during the search optimization calculation process;

[0225] The search optimization solution module is used to perform search optimization solution in combination with the multiple random search points and the multiple candidate search points to increase the probability of jumping out of the local optimal solution and determining the global optimal solution.

[0226] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the aforementioned method embodiment.

[0227] An embodiment of the present invention further provides an electronic device, the device comprising a processor and a memory:

[0228] The memory is used to store the program code and transmit the program code to the processor;

[0229] The processor is used to execute the model predictive control method of the dual active bridge converter according to any embodiment of the present invention according to the instructions in the program code.

[0230] An embodiment of the present invention further provides a computer-readable storage medium, which is used to store program code, and the program code is used to execute the model predictive control method of the dual active bridge converter of any embodiment of the present invention.

[0231] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0232] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0233] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0234] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0235] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0236] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A model predictive control method for a dual active bridge converter, characterized in that: include: Obtaining a predictive control model, an objective cost function, and a moving discrete control set at the end of the dth cycle of the dual active bridge converter, wherein the objective cost function is constructed based on minimum voltage prediction error, minimum voltage change, and minimum current stress as control constraints; The mobile discrete control set includes a plurality of candidate search points, each of which corresponds to a control variable combination; Calculate the output voltage prediction value of each candidate search point at the end of the d+2th cycle according to the prediction control model; For each candidate search point, the control variable combination and the output voltage prediction value are substituted into the target cost function to perform search optimization calculation, and the candidate search point corresponding to the control variable combination that minimizes the target cost function value is used as the optimal operating point at the end of the d+2th cycle; The process of constructing the target cost function includes: The first cost function is constructed with the minimum voltage prediction error as the control constraint. The first cost function is as follows: Where G1 represents the first cost function; V 2,ref represents the target reference value of the output voltage; V2[d+2] represents the predicted value of the output voltage at the end of the d+2th cycle; The second cost function is constructed with the minimum voltage change as the control constraint. The second cost function is as follows: <h2 style=";text-align:left;direction:ltr">G2 = (V2[d] - V2[d+2])<h2 style=";text-align:left;direction:ltr"> 2 Wherein, G2 represents the second cost function; V2[d] represents the output voltage value at the end of the dth cycle; The third cost function is constructed with the minimum current stress as the control constraint. The third cost function is as follows: Among them, G3 represents the third cost function; D and f s They represent the control variable combination corresponding to the current stress (D, f s ) in the external phase angle and switching frequency; D * and They represent minimizing the current peak value, that is, approaching the theoretical minimum value of current stress i peak,1,min The external phase shift angle and switching frequency; According to the first cost function, the second cost function and the third cost function, a target cost function is constructed, and the target cost function is as follows: ct=α1G1+α2G2+α3G3 Among them, ct represents the target cost function; α1, α2, and α3 are all weight factors.

2. The model predictive control method for a dual active bridge converter according to claim 1, characterized in that: The construction process of the predictive control model includes: Constructing a circuit reduction model of a dual active bridge converter, and deriving a capacitor dynamic equation based on the circuit reduction model; Discretizing the capacitor dynamic equation to obtain a current output voltage prediction value at the end of the d+1th cycle; Assuming that the output currents at the end of the dth cycle, the end of the d+1th cycle, and the end of the d+2th cycle are the same, a predictive control model corresponding to the output voltage of the dual active bridge converter at the end of the d+2th cycle is derived based on the current output voltage prediction value.

3. The model predictive control method for a dual active bridge converter according to claim 1, characterized in that: The process of constructing the mobile discrete control set includes: Taking the best control variable combination at the end of the d+1th period predicted at the end of the d-1th period as the center, multiple candidate search points are diffused to form a moving discrete control set.

4. The model predictive control method for a dual active bridge converter according to claim 3, characterized in that: The optimal control variable combination includes the predicted optimal external phase shift angle and the optimal switching frequency at the end of the d+1th cycle. The optimal control variable combination at the end of the d+1th cycle predicted at the end of the d-1th cycle is used as the center, and multiple candidate search points are diffused around to form a mobile discrete control set, including: Obtaining a target reference value of the output voltage and an output voltage value at the end of the d-th cycle, and calculating an adaptive step length of a diffusion search range according to the target reference value of the output voltage and the output voltage value at the end of the d-th cycle; According to the adaptive step length, the first preset basic step length and the first preset adjustment factor, the adaptive step length of the external phase angle is calculated, and the optimal external phase angle is taken as the center, and the external phase angle adaptive step length is used to perform mobile discretization to obtain multiple candidate displacement phase angles; According to the adaptive step length, the second preset basic step length and the second preset adjustment factor, the switching frequency adaptive step length is calculated, and with the optimal switching frequency as the center, the switching frequency adaptive step length is moved and discretized to the surrounding areas to obtain a plurality of candidate switching frequencies; Based on the multiple candidate displacement phase angles and the multiple candidate switching frequencies, multiple candidate search points are determined to form a moving discrete control set corresponding to the end time of the dth period.

5. The model predictive control method for dual active bridge converter according to claim 4, characterized in that: The calculation formula of the adaptive step size is: The calculation formula of the external phase angle adaptive step size is: D adp =D f1 (1+λ1V Δ ) The calculation formula of the switching frequency adaptive step size is: D adf =D f2 (1+λ2V Δ ) Among them, V Δ Represents the adaptive step size, corresponding to the voltage difference between the reference voltage value and the actual output voltage value; V 2,ref represents the target reference value of the output voltage; V2[d] represents the output voltage value at the end of the dth cycle; V m Indicates the preset maximum step size, corresponding to the maximum voltage difference between the reference voltage value and the actual output voltage value; Δ adp Represents the adaptive step size of the external phase angle; Δ f1 represents the first preset basic step length; λ1 represents the first preset adjustment factor; Δ adf Indicates the switching frequency adaptive step size; Δ f2 represents the second preset basic step size; λ2 represents the second preset adjustment factor.

6. The model predictive control method for a dual active bridge converter according to any one of claims 1 to 5, characterized in that: Also includes: In the search optimization calculation process, on the basis of the plurality of candidate search points, a plurality of random search points are additionally added; The multiple random search points and the multiple candidate search points are combined to perform search optimization and solution, so as to increase the probability of escaping from the local optimal solution and determining the global optimal solution.

7. A model predictive control device for a dual active bridge converter, characterized in that: include: A prediction data acquisition module, used to obtain a prediction control model of a dual active bridge converter, a target cost function, and a moving discrete control set at the end of the dth cycle, wherein the target cost function is constructed based on minimum voltage prediction error, minimum voltage change, and minimum current stress as control constraints; The mobile discrete control set includes a plurality of candidate search points, each of which corresponds to a control variable combination; An output voltage prediction value calculation module, used for calculating the output voltage prediction value of each candidate search point at the end of the d+2th cycle according to the prediction control model; A search optimization calculation module, for substituting the control variable combination and the output voltage prediction value into the target cost function for each candidate search point to perform search optimization calculation, and taking the candidate search point corresponding to the control variable combination that minimizes the target cost function value as the optimal operating point at the end of the d+2th cycle; The device also includes: The first cost function construction module is used to construct a first cost function with the minimum voltage prediction error as a control constraint. The first cost function is as follows: G1=(V 2,ref -V2[d+2]) 2 Where G1 represents the first cost function; V 2,ref represents the target reference value of the output voltage; V2[d+2] represents the predicted value of the output voltage at the end of the d+2th cycle; The second cost function construction module is used to construct a second cost function with minimum voltage change as a control constraint. The second cost function is as follows: <h2 style=";text-align:left;direction:ltr">G2 = (V2[d] - V2[d+2])<h2 style=";text-align:left;direction:ltr"> 2 Wherein, G2 represents the second cost function; V2[d] represents the output voltage value at the end of the dth cycle; The third cost function construction module is used to construct a third cost function with the minimum current stress as the control constraint. The third cost function is as follows: Among them, G3 represents the third cost function; D and f s They represent the control variable combination corresponding to the current stress (D, f s ) in the external phase angle and switching frequency; D * and They represent minimizing the current peak value, that is, approaching the theoretical minimum value of current stress i peak,1,min The external phase shift angle and switching frequency; The target cost function construction module is used to construct a target cost function according to the first cost function, the second cost function and the third cost function. The target cost function is as follows: ct=α1G1+α2G2+α3G3 Among them, ct represents the target cost function; α1, α2, and α3 are all weight factors.

8. An electronic device, characterized in that: The device comprises a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the model predictive control method of the dual active bridge converter according to any one of claims 1 to 6 according to the instructions in the program code.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store program code, and the program code is used to execute the model predictive control method of the dual active bridge converter according to any one of claims 1-6.

Images