A power electronic transformer and a method and system for optimizing backflow power of an isolation stage thereof

Abstract

The application discloses a power electronic transformer and an isolation stage backflow power optimization method and system thereof, and the method comprises the following steps: constructing backflow power functions and transmission power calculation models in two working modes; approximating extreme values of the established mathematical models of control objects by using an NAG algorithm to obtain optimal solutions, and obtaining an expression of an internal phase shift ratio D1 of a DAB converter; constructing a deadbeat control model which does not need an additional PI controller compensation according to the relationship between output measured voltages and currents, and then obtaining an expression of an external phase shift ratio D2, collecting voltages and currents of the DAB converter to obtain the external phase shift ratio D2; and realizing adaptive leakage inductance and output capacitance parameters by using a recursive least square method to improve the prediction accuracy of the deadbeat control, and finally sending the obtained optimal internal and external phase shift ratios into an extended phase shift control circuit to realize control on the DAB module. The application can improve the dynamic performance of the system, reduce the backflow power of the system, reduce the system loss, and improve the overall operation efficiency.

Description

Technical Field

[0001] This invention belongs to the field of power electronic converter technology, and more specifically, relates to a power electronic transformer based on parameter adaptive deadbeat predictive control and its isolation stage return power optimization method and system. Background Technology

[0002] With the increasing number of distributed power sources (wind power, hydropower, solar power, tidal power, etc.) and bidirectionally controllable loads (such as energy storage, electric vehicles, etc.) connected to the power grid, the grid's absorption capacity has faced significant challenges. To address new issues such as bidirectional power regulation, AC / DC hybrid network management, and energy storage management in distribution networks, a hybrid AC / DC distribution network with Power Electronic Transformers (PETs) as the interface and management core has been proposed. Leveraging the advantages of PETs—multi-port, multi-functional, flexible expansion, and controllable energy—microgrids based on the PET interface can flexibly manage distributed power sources, electric vehicles, and energy storage systems, thereby supporting the coordinated and optimized operation of the "source-grid-load-storage" system within the energy internet.

[0003] The typical structure of the isolation stage in a power electronic transformer is a dual active bridge (DAB) converter, which offers advantages such as bidirectional energy transfer, high energy density, wide voltage conversion range, and zero-voltage switching of power devices. It has three degrees of freedom in control, providing flexible control and enabling both step-up and step-down conversions. As a key technology in power electronic transformers, the DAB converter largely determines their performance. Therefore, research on high-performance DAB converter technology is of great significance to the development of power electronic transformers. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for optimizing the return current power of a power electronic transformer and its isolation stage based on parameter adaptive deadbeat predictive control. This method can improve the dynamic performance of the system, reduce the return current power of the system, reduce system losses, and improve the overall operating efficiency. It also uses the recursive least squares method to achieve adaptive adjustment of leakage inductance and output capacitance parameters so that the estimated values ​​are closer to the actual values, thereby improving the prediction accuracy of deadbeat control and ultimately improving the robustness of parameter mismatch.

[0005] The first aspect of the present invention provides a method for optimizing the return power of the isolation stage of a power electronic transformer. The optimization method is based on parameter adaptive deadbeat predictive control and includes the following steps:

[0006] Based on the relationship between the inward shift ratio and the outward shift ratio, the operating modes of the DAB converter under EPS control are divided into two types, and the return power function model and transmission power calculation model are constructed for the two operating modes.

[0007] The extreme value of the established return power function model is approximated by the NAG algorithm, and the return power of the DAB converter under EPS control is optimized to obtain the inner shift ratio D1 of the DAB converter.

[0008] Based on the voltage and current relationship on the output side, a deadbeat control model that does not require additional PI controller compensation is constructed by substituting it into the transmission power calculation model. This leads to the expression for the external shift ratio D2, which is used to obtain the external shift ratio D2 by collecting the voltage and current of the DAB converter.

[0009] The recursive least squares method is used to achieve adaptive control of leakage inductance and output capacitance parameters. An optimization model is established to approximate the actual values ​​with the estimated values, and the iterative expression for parameter estimation is obtained. The final optimal inner shift ratio D1 and outer shift ratio D2 are then fed into the EPS control circuit.

[0010] Preferably, the construction of the return power function model and the transmission power calculation model under the two operating modes includes:

[0011] If the inner shift ratio D1 is less than or equal to the outer shift ratio D2, the DAB converter operates in mode 1; if the inner shift ratio D1 is greater than or equal to the outer shift ratio D2, the DAB converter operates in mode 2.

[0012] Establish the primary-side inductor current I of the DAB converter in one cycle under both operating modes. L expression;

[0013] Based on the obtained primary inductance current I L Based on the port voltage of the primary-side full-bridge H1, establish expressions for the return power and transmission power under two operating modes;

[0014] The return power and transmission power are normalized using the maximum transmission power as the power reference value to obtain the return power function and the transmission power calculation model.

[0015] Preferably, the primary-side inductor current I of the DAB converter in one cycle under the two operating modes is established. L The expressions include:

[0016] Using the switching frequency, switching period T, transformer leakage inductance, transformer turns ratio, secondary DC side output voltage, and voltage transfer ratio under two operating modes, establish an expression for the primary inductor current at four times within the period: 0, D1T / 2, D2T / 2, and (D1+D2)T / 2.

[0017] Preferably, the step of normalizing the return power and transmission power using the maximum value of the transmission power as a power reference value to obtain the return power function and transmission power calculation model includes:

[0018] After the transmission power is normalized, the transmission power calculation modulus is obtained in two working modes, characterized only by the inward shift ratio D1 and the outward shift ratio D2.

[0019] After normalization of the return power, we obtain return power function models characterized only by voltage transfer ratio, inward shift ratio D1, and outward shift ratio D2 under the two operating modes.

[0020] Preferably, the approximation of the extreme values ​​of the established return power function model using the NAG algorithm includes:

[0021] Based on the backflow power function model, the backflow power objective function is established by gradient descent, and the gradient expression and optimization iteration expression of the backflow power under EPS control are obtained.

[0022] By incorporating momentum into the gradient expression and the optimization iteration expression, and updating the momentum using future gradients, the NAG backflow power optimization algorithm model is obtained.

[0023] Preferably, the step of adding momentum to the gradient expression and the optimization iteration expression, and updating the momentum using future gradients to obtain the NAG backflow power optimization algorithm model includes:

[0024] The gradient descent is transformed into an equivalent gradient step value after momentum optimization, and the inward shift relative to D1 optimization iteration expression is transformed into the form of iterative addition of the inward shift relative to the initial value and the equivalent gradient step value.

[0025] Preferably, the step of constructing a deadbeat control model that does not require additional PI controller compensation based on the output voltage-current relationship, and thus obtaining the expression for the outward shift ratio D2, includes:

[0026] Establish the expression for the relationship between output voltage and current;

[0027] Based on the established output voltage-current relationship, and taking into account the reference value of the secondary DC-side output voltage, the output current I of the secondary full-bridge H2 is established. O expression;

[0028] Under steady state, the ratio of output power to output current is substituted into the output current I. O The expression yields a comparison of the outward shifts (D2) under both working modes.

[0029] Preferably, the adaptive use of recursive least squares method to achieve the parameters of leakage inductance and output capacitance, and the establishment of an optimization model for approximating the actual values ​​of the estimated values, yields the following iterative expressions for parameter estimation:

[0030] Establish the expressions for the secondary-side output capacitor voltage and leakage inductor current;

[0031] The forward Euler method is used to discretize the equations for the secondary output capacitor voltage and leakage inductor current, resulting in discretized expressions for the secondary output capacitor voltage and leakage inductor current.

[0032] Based on the recursive least squares method, the discretized equations for the secondary output capacitor voltage and leakage inductor current are transformed into matrix form to obtain the estimated values ​​of the output capacitor and leakage inductor.

[0033] Taking into account both the estimated and actual parameter values, a cost function is constructed.

[0034] Minimize the cost function to obtain the iterative form of the recursive least squares method, and control the output optimized phase shift control quantities, inner shift ratio D1 and outer shift ratio D2, according to the converter's operating mode.

[0035] A second aspect of the present invention provides a power electronic transformer, comprising: a DAB converter and a control module, characterized in that the control module performs control of the DAB converter by executing the power electronic transformer isolation stage return power optimization method as described in the first aspect.

[0036] A third aspect of the present invention provides a power electronic transformer isolation stage return power optimization system, comprising: a return power calculation module, a gradient descent optimization module, a deadbeat controller, and an EPS control circuit;

[0037] The return power calculation module is used to calculate the return power based on the sampled electrical quantities of the DAB converter and send it to the gradient descent optimization module.

[0038] The gradient descent optimization module is used to approximate the extreme value of the established return power function model using the NAG algorithm, optimize the return power of the DAB converter under EPS control, and obtain the inner shift ratio D1 of the DAB converter.

[0039] The deadbeat controller is used to construct a deadbeat control model that does not require additional PI controller compensation by substituting the voltage and current relationship on the output side into the transmission power calculation model, thereby obtaining the outward shift ratio D2;

[0040] The gradient descent optimization module and the deadbeat controller send the optimal inward shift ratio D1 and outward shift ratio D2 to the EPS control circuit.

[0041] In summary, compared with the prior art, the power electronic transformer and its isolation stage return power optimization method and system based on parameter adaptive deadbeat predictive control proposed in this invention have the following beneficial effects:

[0042] First, the method for optimizing the return current power of the isolation stage of a power electronic transformer with deadbeat predictive control of the present invention first takes the return current power of the converter as the optimization target. Through the working mode analysis of the converter based on the optimization target, the optimal inner phase shift angle D1 that satisfies the return current power is derived: it can achieve the minimum return current power of the transformer, reduce the power loss of the DAB converter, and achieve the high transmission efficiency of the DAB converter.

[0043] Secondly, this invention enables real-time and rapid adjustment of the converter, improving its dynamic performance and maintaining power transmission stability. By constructing a complete deadbeat control system that minimizes return current power and eliminates the need for an additional PI controller for compensation, the system can quickly adjust to sudden changes in input voltage or load, thus improving stability. Furthermore, the control strategy allows for phase-shift optimization across the entire power range at different voltage turns ratios, ensuring stable converter operation.

[0044] Third, this invention adaptively adjusts the inductor and capacitor parameters of the converter, which can improve the robustness of the converter. Since deadbeat control is highly dependent on parameter values, the recursive least squares method is used to adaptively adjust the leakage inductance and output capacitor parameters, making the estimated values ​​closer to the actual values. This improves the prediction accuracy of deadbeat control and ultimately enhances the robustness of parameter mismatch. Attached Figure Description

[0045] Figure 1 This is a topology diagram of a dual active bridge DC-DC converter in the power electronic transformer isolation stage return power optimization method of the present invention with deadbeat predictive control;

[0046] Figure 2 This is a closed-loop control block diagram of the power electronic transformer isolation stage return power optimization method with deadbeat predictive control according to the present invention.

[0047] Figure 3 This is a flowchart of the process of the method for optimizing the return current power of the isolation stage of a power electronic transformer with deadbeat predictive control according to the present invention.

[0048] Figure 4 This is a simulation diagram of the voltage dynamic performance optimization of the DAB converter under deadbeat predictive control in this invention;

[0049] Figure 5 This is a simulation diagram of the return power optimization of the DAB converter under deadbeat predictive control in this invention. Detailed Implementation

[0050] The following will be combined with the appendix in the embodiments of the present invention. Figure 1 ~Attached Figure 5 The technical solutions, structural features, objectives and effects achieved in the embodiments of the present invention will be described in detail.

[0051] It should be noted that the accompanying drawings are in a very simplified form and use non-precise proportions. They are only used to facilitate and clarify the purpose of illustrating the embodiments of the present invention, and are not intended to limit the implementation conditions of the present invention. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportional relationship, or adjustments to the size should still fall within the scope of the technical content disclosed in the present invention, provided that they do not affect the effects and objectives that the present invention can produce.

[0052] Embodiment 1 of the present invention provides a method for optimizing the return current power of the isolation stage of a power electronic transformer based on deadbeat predictive control, wherein the DAB converter is composed of... Figure 1 The circuit topology shown is implemented.

[0053] Specifically, the DAB converter is a three-level DAB converter, including: a DC power supply, a high-frequency transformer, a primary-side full-bridge converter H1, a secondary-side full-bridge converter H2, a leakage inductance L, a primary-side capacitor C1, and a secondary-side output capacitor C2. The primary-side full-bridge converter H1 includes fully controlled switches Q1, Q2, Q3, and Q4, while the secondary-side full-bridge converter H2 includes fully controlled switches Q5, Q6, Q7, and Q8. Switches Q1, Q2, Q3, and Q4 are all turned on and off with a 50% duty cycle, and switches Q1 and Q2, and Q3 and Q4, are complementary in their conduction. U1 is the primary-side DC input voltage, and U2 is the secondary-side DC output voltage. The turns ratio of the high-frequency transformer is n:1.

[0054] Figure 2 This is a closed-loop control block diagram of the deadbeat predictive control method for optimizing the return current power of the isolation stage of a power electronic transformer according to the present invention; based on the sampled primary DC side input voltage U1, secondary DC side output voltage U2, and output current I2, the per-unit value of the output power P is calculated. * The voltage transfer ratio K is calculated based on the set optimal control algorithm to determine the optimal inner shift ratio D1 for the current state. Then, the outward shift ratio D2 is obtained by the deadbeat controller. The phase shift control of the DAB converter is used to stabilize its output voltage and minimize the return power of the DAB converter when the power supply or load changes.

[0055] Figure 3 This is a flowchart illustrating the workflow of the deadbeat predictive control method for optimizing the return current power of the isolation stage of a power electronic transformer according to the present invention. It is based on the primary-side DC input voltage U1 of the primary-side full-bridge H1 and the primary-side inductor current I. LThe first half of the switching cycle is symmetrical to the second half of the switching cycle. Therefore, the power transmission of the converter is explained by analyzing the inductor current in the first half of the switching cycle. The principle of the second half of the cycle is the same as that of the first half of the cycle.

[0056] like Figure 2 , 3 As shown, the method for optimizing the return power of the isolation stage of a power electronic transformer specifically includes the following steps:

[0057] Step 1: Based on the relationship between the inner and outer shift ratios, the operating modes of the DAB converter under EPS (Extended Phase-Shift Control) are divided into two types, and the return power function and transmission power calculation models for the two operating modes are constructed.

[0058] Preferably, but not limitingly, step 1 specifically includes:

[0059] Step 1.1: Based on the relationship between the inner shift ratio D1 and the outer shift ratio D2, the operating modes of the DAB converter under the extended phase shift control method are divided into two types.

[0060] Specifically, the DAB converter employs phase-shift control. In extended phase-shift control mode, the primary bridge voltage of the transformer is a three-level square wave, and the secondary bridge voltage is a two-level square wave. The phase shift ratio between switching transistors Q1 and Q4 is the inner phase shift ratio D1, and the phase shift ratio between switching transistors Q1 and Q5 is the outer phase shift ratio D2. Changing the magnitude of the outer phase shift ratio D2 changes the direction and magnitude of energy flow. Based on the relationship between the inner and outer phase shift ratios D1 and D2, the DAB converter operates in two modes:

[0061] Mode 1:

[0062] 0≤D1≤D2≤1

[0063] Mode 2:

[0064] 0≤D2≤D1≤1

[0065] In the formula:

[0066] D1 and D2 represent the inward shift ratio and the outward shift ratio, respectively.

[0067] Step 1.2: Based on the operating mode defined in Step 1.1, establish the primary-side inductor current I of the DAB converter within one cycle. L The expressions are as follows:

[0068] Mode 1:

[0069]

[0070] Mode 2:

[0071]

[0072] In the formula:

[0073] I L (t0), I L (t1), I L (t3), I L (t4) is the inductor current in each stage of the switching cycle of the DAB converter, t0=0, t1=D1·T / 2, t3=D2·T / 2, t4=(D1+D2)·T / 2;

[0074] T represents the switching period of the DAB converter;

[0075] f s f represents the switching frequency. s =1 / T;

[0076] L represents the leakage inductance of the high-frequency transformer;

[0077] n represents the turns ratio of the high-frequency transformer;

[0078] U2 represents the secondary DC-side output voltage;

[0079] K represents the voltage transfer ratio.

[0080] Step 1.3, based on the obtained primary-side inductor current I L Based on the port voltage of the primary-side full-bridge H1, establish expressions for the return power and transmission power under two operating modes.

[0081] Specifically, based on the average power calculation formula for the DAB converter under extended phase shift, two expressions for the transmission power under different operating modes are established for subsequent per-unit normalization of the transmission power, namely:

[0082] Mode 1:

[0083]

[0084] Mode 2:

[0085]

[0086] In the formula:

[0087] PEPS and QEPS represent transmission power and return power, respectively.

[0088] U1 and U2 represent the primary side DC input voltage and the secondary side DC output voltage, respectively.

[0089] T represents the switching period of the DAB converter;

[0090] I L (t) represents the primary inductor current;

[0091] D1 and D2 represent the inward shift ratio and the outward shift ratio, respectively;

[0092] L represents the leakage inductance of the high-frequency transformer;

[0093] f s Indicates the switching frequency;

[0094] K represents the voltage transfer ratio.

[0095] Step 1.4: Normalize the return power and the transmission power using the maximum value of the transmission power as the power reference value.

[0096] Specifically, the transmission power and return power are normalized to obtain the per-unit values ​​of the transmission power and return power of the DAB converter in two modes, as follows:

[0097]

[0098] In the formula:

[0099] These represent the per-unit values ​​of transmission power and return power, respectively.

[0100] PEPS and QEPS represent transmission power and return power, respectively, which are the actual transmission power and return power.

[0101] K represents the voltage transfer ratio;

[0102] D1 and D2 represent the inward shift ratio and the outward shift ratio, respectively;

[0103] PN represents the maximum transmission power, and its value is:

[0104]

[0105] In the formula:

[0106] PN represents the maximum transmission power;

[0107] n represents the turns ratio of the high-frequency transformer;

[0108] U1 and U2 represent the primary side DC input voltage and the secondary side DC output voltage, respectively.

[0109] f s Indicates the switching frequency;

[0110] L represents the leakage inductance of the high-frequency transformer.

[0111] Step 2 involves using the NAG (Nesterov Accelerated Gradient) algorithm to approximate the extreme values ​​of the established mathematical model of the controlled object and obtain the optimal solution. This optimizes the return power of the DAB converter under EPS control, reduces the return power, and improves the system transmission efficiency, ultimately yielding the iterative expression for the inner shift ratio of the DAB converter.

[0112] Preferably, but not limitingly, step 2 specifically includes:

[0113] Step 2.1: Based on the principle of gradient descent algorithm, with return power as the objective function, establish the gradient expression and optimization iterative expression for return power under EPS modulation, as shown in the following formula:

[0114]

[0115] In the formula:

[0116] μ is the learning rate;

[0117] D 1(k) The shift is relative to the initial value;

[0118] D 1(k+1) Compared to the iterative update of the inward shift.

[0119] Step 2.2: To reduce the risk of getting trapped in local optima when solving the system's backflow power optimization problem, momentum is added to the original iterative equation, and the momentum is updated using future gradients, forming the Nesterov accelerated gradient descent backflow power optimization algorithm. The original gradient descent expression is then transformed into:

[0120]

[0121] In the formula:

[0122] h is the equivalent gradient step value after adding momentum optimization;

[0123] γ is a hyperparameter of the momentum term, preferably, but not limited to, less than or equal to 0.9.

[0124] Therefore, the iterative expression for the inward shift relative to D1 after adding momentum can be expressed by the following formula:

[0125] D 1(k+1) =D 1(k) +h

[0126] In the formula:

[0127] h is the equivalent gradient step value after adding momentum optimization;

[0128] D 1(k)The shift is relative to the initial value;

[0129] D 1(k+1) Compared to the iterative update of the inward shift.

[0130] One of the most prominent features of this invention is that when D1 is updated, not only the current gradient value is considered, but future gradients are also used to update momentum, resulting in a faster convergence speed compared to the traditional momentum gradient descent algorithm. Regarding the momentum gradient, when the gradient and momentum are in the same direction, the momentum increases, improving the solution speed of D1; when the gradient and momentum are in opposite directions, the momentum decreases, thereby suppressing system oscillations and overshoot.

[0131] Step 3: Based on the voltage and current relationship on the output side, in order to improve the dynamic response capability and meet the requirements of fast output voltage tracking and minimum return current power, a deadbeat control model that does not require additional PI controller compensation is constructed, thereby obtaining the expression for the external shift ratio D2. The voltage and current of the DAB converter are collected to obtain the external shift ratio D2.

[0132] Preferably, but not limitingly, step 3 specifically includes:

[0133] Step 3.1: Establish the output-side voltage-current relationship expression, expressed by the following formula:

[0134]

[0135] In the formula:

[0136] C2 represents the capacitance value of the secondary output capacitor;

[0137] U2 represents the secondary DC-side output voltage;

[0138] I O This indicates the output current of the secondary full-bridge H2, such as... Figure 1 As shown;

[0139] I2 represents the output current of the DAB converter.

[0140] Step 3.2: Based on the output voltage-current relationship established in Step 3.1, and taking into account the reference value of the secondary DC-side output voltage, establish the output current I of the secondary full-bridge H2. O The expression. Specifically, if the secondary DC-side output voltage U2 is controlled to its reference value (U2 = U... ref If the output current I of the secondary full-bridge H2 is... O The value can be obtained as follows:

[0141] I O =I² + C²f s (U2-U ref )

[0142] In the formula:

[0143] I O This indicates the output current of the secondary full-bridge H2;

[0144] I2 represents the output current of the DAB converter;

[0145] C2 represents the capacitance value of the secondary output capacitor;

[0146] f s Indicates the switching frequency;

[0147] U2 represents the secondary DC-side output voltage;

[0148] U ref This indicates the reference value for the DC-side output voltage on the secondary side.

[0149] Step 3.3: Under the assumption of steady-state conditions, by I O Substituting P / U2 into the output current I of the secondary full-bridge H2 established in step 3.2, O The expression yields the following values ​​for the outward shift comparison D2 under the two working modes:

[0150] Mode 1:

[0151] When 0 ≤ D1 ≤ D2 ≤ 1, the value of the outward shift relative to D2 can be derived as follows:

[0152]

[0153] Mode 2:

[0154] Similarly, when 0 ≤ D2 ≤ D1 ≤ 1, the value of the outward shift relative to D2 can be derived as follows:

[0155]

[0156] In the formula:

[0157] D1 and D2 represent the inward shift ratio and the outward shift ratio, respectively;

[0158] n represents the turns ratio of the high-frequency transformer;

[0159] U1 and U2 represent the primary DC-side input voltage and the secondary DC-side output voltage, respectively. ref This indicates the reference value for the DC-side output voltage on the secondary side;

[0160] f s Indicates the switching frequency;

[0161] L represents the sum of the leakage inductance and auxiliary inductance of the high-frequency transformer;

[0162] C2 represents the capacitance value of the secondary output capacitor.

[0163] I2 represents the output current of the DAB converter.

[0164] In the real-time control of DAB converters, the optimal phase shift angle under different power levels and voltage ratios can be dynamically analyzed, and the control parameters can be adjusted based on the real-time operating conditions to achieve adaptive optimization control of the converter's operating state.

[0165] Step 4: The recursive least squares method is used to achieve adaptive control of leakage inductance and output capacitance parameters, thereby improving the prediction accuracy of deadbeat control and making the estimated value closer to the actual value. The iterative expression of parameter estimation is obtained, and the final optimal internal and external phase shift ratio is sent to the extended phase shift control circuit to realize the control of the DAB converter.

[0166] It is worth noting that for deadbeat control, the prediction accuracy is related to the accuracy of the transformer leakage inductance and the secondary output capacitor. In other words, the outward shift ratio D2 obtained by deadbeat control is closely related to the values ​​of leakage inductance and output capacitor. Therefore, this invention uses the recursive least squares method to achieve parameter adaptation and thus improve the prediction accuracy of deadbeat control.

[0167] Preferably, but not limitingly, step 4 specifically includes:

[0168] Step 4.1: Establish the expressions for the secondary output capacitor voltage and leakage inductor current, as shown in the following formula:

[0169]

[0170] In the formula:

[0171] C2 represents the capacitance value of the secondary output capacitor;

[0172] U1 and U2 represent the primary side DC input voltage and the secondary side DC output voltage, respectively.

[0173] I O This indicates the output current of the secondary full-bridge H2;

[0174] I2 represents the output current of the DAB converter;

[0175] L represents the sum of the leakage inductance and auxiliary inductance of the high-frequency transformer;

[0176] n represents the turns ratio of the high-frequency transformer.

[0177] Step 4.2: Discretize the equations for the secondary-side output capacitor voltage and leakage inductor current using the forward Euler method, and assume that the output current and input voltage remain approximately constant during the sampling period. The discretized expressions for the secondary-side output capacitor voltage and leakage inductor current can be obtained as follows:

[0178]

[0179] In the formula:

[0180] I 2[k] This represents the k-th discrete value of the output current of the DAB converter after discretization.

[0181] I O[k-1] This represents the (k-1)th discrete value of the output current of the secondary full-bridge H2 after discretization.

[0182] U 2[k] U 2[k-1] These represent the kth and (k-1th)th discrete values ​​of the secondary DC-side output voltage after discretization, respectively.

[0183] U 1[k] This represents the k-th discrete value of the primary DC-side input voltage after discretization.

[0184] I L[k] I L[k-1] These represent the k-th and (k-1)-th discrete values ​​of the primary inductor current after discretization, respectively.

[0185] C2 represents the capacitance value of the secondary output capacitor;

[0186] T represents the switching period of the DAB converter;

[0187] n represents the turns ratio of the high-frequency transformer;

[0188] L represents the leakage inductance of the high-frequency transformer.

[0189] Step 4.3: Based on the principle of recursive least squares, the discretized equations for the secondary output capacitor voltage and leakage inductor current are transformed into matrix form, yielding estimated values ​​for the output capacitor and leakage inductor, expressed by the following formulas:

[0190]

[0191] In the formula:

[0192] This represents the matrix of estimated values ​​of U1 and I2 at time k;

[0193] This represents the estimated output current of the DAB converter at time k.

[0194] This represents the estimated value of the primary side DC input voltage at time k;

[0195] Represents the measurement matrix at time k;

[0196] This represents the estimated parameter matrix at time k-1.

[0197] I O[k-1] This represents the (k-1)th discrete value of the output current of the secondary full-bridge H2 after discretization.

[0198] U 2[k] U 2[k-1] These represent the kth and (k-1th)th discrete values ​​of the secondary DC-side output voltage after discretization, respectively.

[0199] I L[k] I L[k-1] These represent the k-th and (k-1)-th discrete values ​​of the primary inductor current after discretization, respectively.

[0200] C2 represents the capacitance value of the secondary output capacitor;

[0201] T represents the switching period of the DAB converter;

[0202] n represents the turns ratio of the high-frequency transformer;

[0203] L represents the leakage inductance of the high-frequency transformer.

[0204] Step 4.4: To make the parameter estimates closer to the actual values, construct the cost function as follows:

[0205]

[0206] In the formula:

[0207] J represents the cost function;

[0208] λ represents the forgetting factor, which is used to enable the model to better adapt to changes in the current data; preferably, but not limited to, the forgetting factor λ is greater than 0.9 and does not exceed 1.

[0209] y [k] , Let U1 and I2 represent the actual and estimated values ​​of the matrix at time k, respectively.

[0210] Step 4.5: Finally, minimize the cost function J, that is, minimize the cost function J with respect to... The partial derivative is 0, thus the iterative form of the recursive least squares method can be obtained:

[0211] Calculate the gain:

[0212]

[0213] Estimated value update:

[0214]

[0215] Covariance update:

[0216]

[0217] In the formula:

[0218] K [k] This represents the gain matrix of the estimator at time k.

[0219] P [k] P [k-1] Let represent the error covariance matrices at time k and time (k-1), respectively.

[0220] Represents the measurement matrix at time k;

[0221] Let represent the estimated parameter matrices at time k and time (k-1), respectively;

[0222] y [k] , Let U1 and I2 represent the actual and estimated values ​​of the matrix at time k, respectively.

[0223] λ represents the forgetting factor.

[0224] Based on the converter's operating mode, the output optimization phase shift control quantities are controlled by the inner shift ratio D1 and the outer shift ratio D2. These are then used to generate drive signals via a PWM pulse width generator, which in turn control the turn-off of eight switching transistors in the primary-side full-bridge H1 and the secondary-side full-bridge H2. Through the above control method, the optimization of return power under the fast dynamic response of the converter output can be achieved.

[0225] Embodiment 2 of the present invention provides a power electronic transformer, including a DAB converter and a control module. The control module executes the power electronic transformer isolation stage return power optimization method as described in Embodiment 1 to control the DAB converter.

[0226] like Figure 2 As shown, Embodiment 3 of the present invention provides a power electronic transformer isolation stage return power optimization system, including: a return power calculation module, a gradient descent optimization module, a deadbeat controller, and an EPS control circuit;

[0227] The return power calculation module is used to calculate the return power based on the sampled electrical quantities of the DAB converter and send it to the gradient descent optimization module.

[0228] The gradient descent optimization module is used to approximate the extreme value of the established return power function model using the NAG algorithm, optimize the return power of the DAB converter under EPS control, and obtain the inner shift ratio D1 of the DAB converter.

[0229] The deadbeat controller is used to construct a deadbeat control model that does not require additional PI controller compensation by substituting the output side voltage and current relationship into the transmission power calculation model, thereby obtaining the outward shift ratio D2;

[0230] The gradient descent optimization module and the deadbeat controller send the optimal inward shift ratio D1 and outward shift ratio D2 to the EPS control circuit.

[0231] Embodiment 4 of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded onto the processor, it implements the power electronic transformer isolation stage return power optimization method based on deadbeat predictive control as described in Embodiment 1.

[0232] Embodiment 5 of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method for optimizing the return current power of the isolation stage of a power electronic transformer based on deadbeat predictive control as described in Embodiment 1.

[0233] To more clearly illustrate the outstanding substantive features of this invention and the significant progress it brings to the prior art, an application example of implementing this invention is described below.

[0234] Finally, a simulation model of the DAB converter was built in Simulink in MATLAB. The converter parameters were designed as follows: the input voltage was set to 500V, the output reference voltage was set to 200V, a 30Ω resistor was used to replace the output power supply, the transformer ratio n = 2:1, the leakage inductance L = 85.6uH, the switching frequency f = 10kHz, the input capacitor C1 = 2000uF, and the output capacitor C2 = 1300uF. Figure 4 and Figure 5 The simulation results show that, due to the deadbeat control, the system stability is improved when subjected to input voltage or output load disturbances, and the system's output voltage can be dynamically adjusted quickly. Furthermore, by incorporating a return current optimization algorithm, the system's return current power is significantly reduced under extended phase-shift control, and the system's steady-state performance is further improved, greatly enhancing the efficiency of the DAB converter.

[0235] It is worth noting that some technologies using AI-based controller parameter tuning require a large amount of data and involve complex processes, or they cannot iteratively update the obtained shift ratio. Direct power control is used to improve dynamic response, with compensation generated by a PI controller. However, the integral stage increases the settling time and does not consider the impact of parameter mismatch on output voltage control. As one of the key features of this invention, the return power optimization employs an accelerated Nesterov gradient descent algorithm. By adding a momentum component to iterate the gradient step size and updating momentum using future gradients, it achieves fast convergence and suppresses oscillations. Deadbeat control is used to quickly bring the voltage to the reference value. For parameter mismatch issues, a sampling recursive least squares method is used for adaptive parameter adjustment, improving the system's dynamic response and robustness. The corresponding beneficial effects include at least: fast output voltage dynamic response and reduction of return power.

[0236] It is worth noting that in the embodiments of the present invention, "steps + numbers" is only an expression for clearly describing the specific implementation method of the power electronic transformer isolation stage return power optimization method based on deadbeat predictive control, and is not an absolute restriction on the order of each step. Under the guidance of the core concept of the present invention, changing the order of these steps to obtain the same or similar technical effects all fall within the scope of the present invention.

[0237] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for optimizing the return current power of the isolation stage of a power electronic transformer, characterized in that, The optimization method is based on parameter adaptive deadbeat predictive control and includes the following steps: Based on the relationship between the inward shift ratio and the outward shift ratio, the operating modes of the DAB converter under EPS control are divided into two types, and the return power function model and transmission power calculation model are constructed for the two operating modes. The NAG algorithm is used to approximate the extremum of the established backflow power function model, including: based on the backflow power function model, establishing the backflow power objective function by gradient descent, obtaining the gradient expression and optimization iteration expression of the backflow power under EPS control; adding momentum to the gradient expression and optimization iteration expression, updating the momentum by using future gradients, and obtaining the NAG backflow power optimization algorithm model. Optimize the return power of the DAB converter under EPS control to obtain the inward shift ratio of the DAB converter. D 1; Based on the output voltage-current relationship, a deadbeat control model that does not require additional PI controller compensation is constructed by substituting it into the transmission power calculation model, thereby obtaining the outward shift ratio. D The expression in 2 is used to obtain the external shift ratio by acquiring the voltage and current of the DAB converter. D 2. This includes: establishing the output-side voltage-current relationship expression; based on the established output-side voltage-current relationship, and taking into account the reference value of the secondary-side DC-side output voltage, establishing a secondary-side full-bridge... H 2 Output Current I O Expression; Under steady state, substitute the ratio of output power to output current into the output current. I O The expression yields a comparison of the outward shift between the two working modes. D 2; A recursive least squares method is used to achieve adaptive operation of leakage inductance and output capacitance parameters. An optimization model is established to approximate the actual values ​​with the estimated values, resulting in iterative expressions for parameter estimation. The final optimal shift is then compared with... D 1. Compared to relocation D 2. The signal is sent to the EPS control circuit.

2. The method for optimizing the return current power of the isolation stage of a power electronic transformer according to claim 1, characterized in that: The construction of the return power function model and transmission power calculation model under the two working modes includes: If the inward shift is compared to D 1 less than or equal to the outward shift ratio D 2. Then the DAB converter operates in mode 1. If the ratio is shifted inward... D 1 is greater than or equal to the outward shift ratio D 2. Therefore, the operating mode of the DAB converter is mode 2. Establish the primary-side inductor current of the DAB converter in one cycle under two operating modes. I L expression; Based on the obtained primary inductance current I L Hoharabe Bridge H Given the port voltage of 1, establish expressions for the return power and transmission power under two operating modes; The return power and transmission power are normalized using the maximum transmission power as the power reference value to obtain the return power function and the transmission power calculation model.

3. The method for optimizing the return current power of the isolation stage of a power electronic transformer according to claim 2, characterized in that: The primary inductor current of the DAB converter in one cycle is established under two operating modes. I L The expressions include: Based on switching frequency and switching cycle T The transformer's leakage inductance, transformer turns ratio, secondary DC-side output voltage, and voltage transfer ratio are established within a cycle at time 0 under two operating modes. D 1 T / 2 time, D 2 T / 2 and ( D 1+ D 2) T The expression for the primary inductor current at these four moments: / 2.

4. A method for optimizing the return current power of the isolation stage of a power electronic transformer according to claim 2 or 3, characterized in that: The step of normalizing the return power and transmission power using the maximum transmission power as the power reference value to obtain the return power function and transmission power calculation model includes: After per-unit processing of transmission power, only the inward shift is compared between the two operating modes. D 1. Compared to relocation D 2. Transmission power calculation model; After per-unit processing of the return power, only the voltage transfer ratio and inward shift are compared in the two operating modes. D 1. Compared to relocation D 2. The return power function model.

5. The method for optimizing the return current power of the isolation stage of a power electronic transformer according to claim 1, characterized in that: The step of adding momentum to the gradient expression and the optimization iteration expression, and updating the momentum using future gradients to obtain the NAG backflow power optimization algorithm model includes: The gradient descent step size is reduced to the equivalent gradient step size after momentum optimization, and the inward shift is compared to... D 1. The optimized iterative expression is transformed into the form of adding the equivalent gradient step numerical iterations of the inner shift compared to the initial value and the value.

6. The method for optimizing the return current power of the isolation stage of a power electronic transformer according to claim 1, characterized in that: The recursive least squares method is used to achieve adaptive control of leakage inductance and output capacitance parameters. An optimization model is established to approximate the actual values ​​with the estimated values, resulting in iterative expressions for parameter estimation, including: Establish the expressions for the secondary-side output capacitor voltage and leakage inductor current; The forward Euler method is used to discretize the equations for the secondary output capacitor voltage and leakage inductor current, resulting in discretized expressions for the secondary output capacitor voltage and leakage inductor current. Based on the recursive least squares method, the discretized equations for the secondary output capacitor voltage and leakage inductor current are transformed into matrix form to obtain the estimated values ​​of the output capacitor and leakage inductor. Taking into account both the estimated and actual parameter values, a cost function is constructed. Minimizing the cost function yields the iterative form of the recursive least squares method. The phase shift control quantity for output optimization is then adjusted based on the converter's operating mode to optimize the phase shift ratio. D 1. Compared to relocation D 2.

7. A power electronic transformer, comprising: The DAB converter and control module are characterized in that the control module executes the power electronic transformer isolation stage return power optimization method as described in any one of claims 1 to 6 to control the DAB converter.

8. A power electronic transformer isolation stage return power optimization system, executing the power electronic transformer isolation stage return power optimization method as described in any one of claims 1 to 6, characterized in that, include: Return power calculation module, gradient descent optimization module, deadbeat controller and EPS control circuit; The return power calculation module is used to calculate the return power based on the sampled electrical quantities of the DAB converter and send it to the gradient descent optimization module. The gradient descent optimization module is used to approximate the extremum of the established return power function model using the NAG algorithm, optimize the return power of the DAB converter under EPS control, and obtain the inward shift ratio of the DAB converter. D 1; The deadbeat controller is used to construct a deadbeat control model that does not require additional PI controller compensation by substituting the output-side voltage-current relationship into the transmission power calculation model, thereby obtaining the outward shift ratio. D 2; The gradient descent optimization module and the deadbeat controller will shift the optimal inward position compared to... D 1. Compared to relocation D 2. The signal is sent to the EPS control circuit.

Images