Online co-optimization method and system based on reflux power and soft switch constraint

Abstract

The application discloses an online collaborative optimization method and system based on reflux power and soft switch constraint, mainly solves the problems of reflux power increase caused by transient power and equivalent voltage fluctuation in the existing power frequency cycle and hard switching caused by crossing the soft switch boundary when refluxing. The scheme comprises the following steps: sampling the real-time state variables running in the dual active bridge micro-inverter circuit; calculating the internal and external phase displacement and the objective function based on the sampled photovoltaic side voltage, photovoltaic side current, leakage inductance current, bridge arm equivalent voltage and grid side voltage; outputting the soft switch judgment signal of the bridge arm commutation current at the commutation event k moment according to the judgment signal; based on the judgment signal, the optimization variables are iteratively updated online to obtain the optimal optimization variables; based on the optimal optimization variables, the corrected internal and external phase displacement are calculated, and the driving signal of the switching tube is outputted to realize the online collaborative optimization of the reflux power and the soft switch constraint of the dual active bridge micro-inverter. The application can reduce the reflux under all working conditions and meet the soft switch, and can be used for reflux power suppression of the dual active bridge converter.

Description

Technical Field

[0001] This invention belongs to the field of power electronics technology, and in particular relates to an online collaborative optimization method and system for return power and soft switching constraints, which can be used for return power suppression in dual active bridge converters. Technical Background

[0002] In distributed photovoltaic (PV) systems, microinverters are used to convert the direct current (DC) output from individual PV modules into grid-connectable alternating current (AC). Compared to string-type systems, microinverters enable module-level control, reducing power generation losses due to individual module mismatch and improving system safety and maintainability. To achieve both electrical isolation and high power density within a small footprint, microinverters often employ a dual active bridge (DAB) high-frequency isolation structure and utilize phase-shift modulation for energy transfer; this structure has the potential to achieve zero-voltage switching (ZVS) within a certain operating range. However, during grid-connected operation, as the grid voltage (Vg) phase continuously changes within the power frequency cycle, the instantaneous power on the grid side also fluctuates periodically, causing the operating conditions to change continuously with the power frequency cycle. In other words, the system does not operate at a single steady-state point but continuously sweeps through different phases and power points within a single power frequency cycle. As a result, the equivalent voltage relationship and leakage inductance branch current of the dual active bridge DAB converter change continuously within the cycle, easily generating more significant return power Pcir in certain phase intervals. This return power manifests as energy being exchanged back and forth within the converter, contributing a limited amount to the external active power output. This return power Pcir increases device current stress and increases conduction and magnetic device losses. It also alters the direction and amplitude of the commutation current, making soft-switching conditions more easily disrupted, thus causing the system to more frequently touch the critical boundaries of zero-voltage turn-on (ZVS) / non-zero-voltage turn-on.

[0003] Patent document CN201910500839.4 discloses a method for optimizing the return current power of a dual active bridge DAB converter. It first collects the actual output voltage Vo and generates the desired normalized transfer power P* by calculating the deviation between Vo and the desired output voltage Vo*. Then, based on P* and the voltage conversion ratio M, it calculates the inner shift ratio D1 and the outer shift ratio D2 to drive the switches of the primary and secondary full-bridge circuits, achieving zero-voltage turn-on (ZVS) or zero-current turn-on (ZCS). While this method minimizes return current power while ensuring soft switching, thus improving efficiency and maintaining voltage regulation accuracy under wide loads, the mapping between the desired normalized transfer power P* and the voltage conversion ratio M and the inner and outer shift ratios D1 and D2 usually relies on table lookups. Therefore, when parameters drift or are inaccurately estimated, the return current power may not be truly minimized, or even the soft-switching margin may decrease.

[0004] Patent application CN202511534729.1 discloses a control method for a dual active bridge (DAB). This method first collects output electrical parameters to calculate the real-time power P, and then calculates the critical power Pcr based on the voltage transfer ratio M. Next, it compares the real-time power P with the critical power Pcr: when P ≤ Pcr, a minimum return current power strategy is used to calculate the inner shift ratio D1 and the outer shift ratio D2; when P > Pcr, a synergistic optimization of return current power and current stress is adopted, using a weighted objective function and iteratively employing the momentum method to find the optimal inner shift ratio D1 and the outer shift ratio D2. Finally, the optimal inner shift ratio D1 and the outer shift ratio D2 are used to generate a PWM to drive the switching transistors in the circuit, thereby reducing return current power and current stress, and improving efficiency and reliability under wide operating conditions. However, this method suffers from critical point drift when there are deviations in leakage inductance Lk, dead time td, device Coss, or voltage measurement. This can lead to incorrect segment selection and a decrease in optimization effectiveness.

[0005] In summary, the shortcomings of existing methods are: the return power Pcir is relatively large, the optimal point changes with the power level and parameter drift, the zero-voltage turn-on (ZVS) boundary is difficult to determine reliably, and the solution capability is still limited. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of the prior art by proposing an online collaborative optimization method and system for return power and soft-switching constraints, so as to reduce the magnitude of return power Pcir and meet the constraints of soft switching; avoid critical point drift caused by parameter deviation, improve the optimization effect on the magnitude of return power, and reduce current stress.

[0007] The technical approach to achieving the objective of this invention is as follows: under the premise of hard constraints on soft-switching ZVS, an online minimization layer is introduced to optimize the variables. Online iteration without lookup table is performed to reduce the objective function J, suppress critical point drift caused by parameter deviation, and minimize the return power Pcir.

[0008] Based on the above ideas, the technical solution of the present invention includes:

[0009] 1. An online collaborative optimization method based on return power and soft-switching constraints, characterized in that it includes:

[0010] S1) Sample the real-time state variables of the dual active bridge micro-inverter circuit, including: photovoltaic side voltage Vpv, DC side voltage Vdc, photovoltaic side current ipv, grid side voltage Vg, grid side phase θ, bridge arm commutation current ileg,k(t), leakage inductance current iLk(t) at commutation event k, and bridge arm equivalent voltage Veq(t);

[0011] S2) Based on the sampled photovoltaic side voltage Vpv, photovoltaic side current ipv, leakage inductance current iLk(t), bridge arm equivalent voltage Veq(t), grid side voltage Vg and grid side phase θ, calculate the inner phase shift D1, the outer phase shift D2 and the objective function J;

[0012] S3) Calculate the actual current Ik based on the sampled bridge arm commutator current ileg,k(t);

[0013] S4) Based on the sign of Ik, output the soft-switching judgment signal at time k of the commutation event, and perform online iterative updates on the value of the optimization variable ξ at the current time based on the soft-switching judgment signal to obtain the optimal optimization variable ξ*.

[0014] S5) Based on the updated optimal optimization variable ξ*, calculate the corrected inner phase shift D1' and outer phase shift D2', and output the drive signal of the switching transistor through the phase shift modulator to realize the online collaborative optimization method of return power and soft switching constraint.

[0015] Furthermore, in step (S5), based on the corrected inner phase shift D1', outer phase shift D2', and switching frequency fs, the driving signal for the switching transistor is output through the phase-shift modulator, which includes:

[0016] (S5a) Based on the optimal optimization variable ξ* and the soft switching decision signal ΔZVS,k, calculate the corrected inner phase shift D1' and outer phase shift D2';

[0017] (S5b) Based on the key signals such as the corrected inner phase shift D1', outer phase shift D2', network measurement phase θ and switching frequency fs, a square wave pulse with phase shift control is generated by the phase shift modulator to drive the switching transistor in the dual active bridge micro inverter circuit to work.

[0018] 2. An online collaborative optimization system based on return power and soft-switching constraints, characterized in that it comprises:

[0019] The feedforward control module is used to generate the inner phase shift quantity D1 and the outer phase shift quantity D2 required for phase shift control based on the measured photovoltaic side voltage Vpv, grid side voltage Vg, grid side phase θ, grid side current ig and the set grid side current reference value Im,ref. These serve as the basic instructions for subsequent soft switching determination and phase shift modulation.

[0020] The soft-switching online determination module is used to sample the arm current ileg,k(t) in real time at the commutation event k, calculate the soft-switching related indicators based on the actual current Ik, and output the soft-switching determination signal.

[0021] The online optimization module for return power is used to iteratively update the optimization variable ξ online based on the soft-switching state output by the online soft-switching determination module, obtain the optimal optimization variable ξ∗, and calculate the corrected inner phase shift D1' and outer phase shift D2'.

[0022] The phase-shift modulator module is used to generate a square wave pulse with a fixed duty cycle of 50% for each switch based on the inner phase shift D1', the outer phase shift D2' and the switching frequency fs. The start time of each pulse is uniquely determined by the corresponding critical action point, and its turn-off time is automatically derived by delaying the start time by half a switching cycle.

[0023] Compared with the prior art, the present invention has the following advantages:

[0024] Firstly, this invention improves the stability of zero-voltage turn-on (ZVS) boundary determination by directly determining the positive or negative value of the actual current Ik online;

[0025] Secondly, since the present invention directly calculates the corrected inner phase shift quantity D1' and outer phase shift quantity D2' to output the drive signal of the switching transistor, instead of relying on table lookup, it avoids parameter drift or inaccurate estimation, and can truly minimize the return current power. Attached Figure Description

[0026] Figure 1 The overall flowchart for the online collaborative optimization method based on return power and soft switching constraints of this invention is shown below.

[0027] Figure 2 This is a flowchart of the online iterative update calculation in the method of the present invention;

[0028] Figure 3 This is a sub-flowchart of the method for outputting the driving signal in this invention;

[0029] Figure 4 This is a block diagram of the online collaborative optimization system based on the return power and soft switching constraints of the dual active bridge micro-inverter according to the present invention. Detailed Implementation

[0030] The embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0031] Example 1: An online collaborative optimization method based on return power and soft switching constraints.

[0032] This example is implemented based on a dual active bridge micro-inverter circuit, which includes three parts: DC side, transformer, and AC side.

[0033] The DC side is a primary-side DC full-bridge circuit, which consists of four switching transistors G1-G4. This full-bridge can generate a three-level high-frequency square wave voltage Vp with an adjustable duty cycle. Its value range includes +Vdc, 0, -Vdc, i.e., Vp={+Vdc,0, -Vdc}, which is used to achieve preliminary demodulation of the input energy.

[0034] The transformer includes a high-frequency transformer and an equivalent leakage inductance. The high-frequency transformer is connected at the midpoint of the primary and secondary circuits and uses its secondary equivalent leakage inductance Lk as an intermediate energy storage medium for energy transmission. The equivalent leakage inductance Lk stores energy using the voltage difference between the primary voltage Vp and the secondary voltage Vs.

[0035] The AC side includes a secondary-side half-bridge frequency converter circuit and the mains voltage. This secondary-side half-bridge consists of four switches G5-G8 and two capacitors C1 and C2. The fifth switch G5 and the sixth switch G6, and the seventh switch G7 and the eighth switch G8, respectively form two sets of bidirectional switch pairs. This secondary-side half-bridge can generate a two-level square wave voltage Vs with a fixed 50% duty cycle, whose value range includes +Vg / 2 and -Vg / 2, i.e., Vs = {+Vg / 2, -Vg / 2}.

[0036] Reference Figure 1 The implementation steps for this example are as follows:

[0037] Step 1: Sample the real-time state variables of the dual active bridge micro-inverter circuit.

[0038] This example uses an ideal voltage measurement unit, current measurement unit, and phase-locked loop (PLL) unit provided by simulation software to sample the key state variables of the dual active bridge micro-inverter circuit in real time. The ideal voltage measurement unit is connected in parallel across the measured signal with high input impedance, its measurement port bridging the target node and the reference node, outputting a continuous-time voltage signal V(t). The ideal current measurement unit is connected in series with the branch of the measured signal with zero series impedance (i.e., zero equivalent voltage drop), outputting a continuous-time current signal i(t). The PLL unit obtains the grid-side phase output by inputting the grid-side voltage. This unit first conditions and discretizes the grid-side voltage Vg before inputting it into a second-order generalized integrator PLL (SOGI-PLL) to generate two mutually orthogonal vd and vq signals. The vq signals are then proportionally and integrally adjusted to obtain a frequency correction, which is added to the rated grid-side angular frequency w0 to obtain an estimated angular frequency. The phase is updated by integrating this estimated angular frequency, achieving continuous tracking of the grid-side phase through this closed-loop process.

[0039] The parameters that need to be sampled include: photovoltaic side voltage Vpv, DC side voltage Vdc, grid side voltage Vg, bridge arm equivalent voltage Veq(t), photovoltaic side current ipv, leakage inductance current iLk(t), bridge arm commutation current ileg,k(t) at commutation event k, and grid side phase θ. The sampling process for each parameter is as follows:

[0040] The photovoltaic-side voltage Vpv, DC-side voltage Vdc, grid-side voltage Vg, and bridge arm equivalent voltage Viq(t) are measured by a parallel ideal voltage measurement unit.

[0041] The photovoltaic side current ipv, leakage inductance current iLk(t) and bridge arm commutation current ileg,k(t) at commutation event k are measured by a series ideal current measurement unit.

[0042] The grid-side phase θ is obtained by real-time processing of the sampled grid-side voltage Vg using a second-order generalized integrator phase-locked loop (SOGI-PLL) unit in software.

[0043] Step 2: Calculate the internal phase shift D1 and external phase shift D2 used by each switch in the control circuit.

[0044] 2.1) Based on the sampled grid-side voltage Vg, DC-side voltage Vdc, and grid-side phase θ, calculate the internal phase shift D1 used by each switch:

[0045] ,

[0046] Where n is the turns ratio of the secondary winding to the primary winding of the transformer, and its value is set to 4;

[0047] 2.2) Based on the sampled grid-side voltage Vg, photovoltaic-side voltage Vpv, and leakage inductance value Lk of the transformer secondary side, calculate the external phase shift D2 used by each switch:

[0048] ,

[0049] Wherein, sgn(Vg) is the sign function of the grid-side voltage Vg, and its value is obtained according to the positive or negative sign of the grid-side voltage Vg, that is, sgn(Vg)∈{+1,-1}; fs is the switching frequency, and its value is set to 20kHz; Im,ref is the reference value of the grid-side current, and its value is set to Im,ref=0.96sin(100πt).

[0050] Step 3, set the objective function J.

[0051] To minimize the return current power, an objective function J needs to be defined, and the implementation steps include the following:

[0052] 3.1) Calculate the power function value P1 based on the sampled leakage inductance current iLk(t) and bridge arm equivalent voltage Veq(t):

[0053] ,

[0054] Where w1 is the power weighting coefficient, and its value is set to 0.8;

[0055] 3.2) Calculate the effective value of the inductor current P2 based on the sampled leakage inductance current iLk(t):

[0056] ,

[0057] Where w2 is the weighting coefficient of the effective value of the inductor current, and its value is set to 0.2;

[0058] 3.4) Based on the power function value P1 and the effective value of the inductor current P2 mentioned above, set the target function J:

[0059] .

[0060] Step 4: Calculate the actual current Ik, and output the soft switching judgment signal at time k of commutation event based on the sign of Ik.

[0061] 4.1) Calculate the actual current Ik based on the sampled commutation current ileg,k(t) of the bridge arm at time k:

[0062] ,

[0063] Where sk∈{+1,-1} represents the current direction required at time k of the commutation event;

[0064] 4.2) Based on the sign of Ik, output the soft-switching judgment signal at time k of the commutation event:

[0065] The soft-switching state refers to a state in which the voltage and current of the switching transistor do not overlap during the transition from turn-on to turn-off in a circuit. If they overlap, it is called hard switching. To achieve soft switching of a switching device, the voltage across its terminals must be reduced to 0 before turn-on or the current flowing through it must be reduced to 0 before turn-off. Therefore, it is mainly divided into zero-voltage turn-on and zero-current turn-off. In zero-voltage turn-on, the voltage across the switch is 0 before turn-on; in zero-current turn-off, the current flowing through the switch is 0 before turn-off. The soft-switching signal is determined based on the sign of the actual current Ik.

[0066] If Ik < 0, then the output soft-switching decision signal ΔZVS,k is 1;

[0067] Otherwise, the output soft-switching judgment signal ΔZVS,k is 0.

[0068] Step 5: Based on the soft switching determination signal at time k of the commutation event, the value of the current optimization variable ξ is updated online to obtain the optimal optimization variable ξ*.

[0069] The online iterative calculation refers to the controller updating the calculation results once according to a preset control cycle or commutation event cycle during operation, and repeating this process cyclically to gradually approach the optimal control parameters. Specifically, the controller first collects the current operating state variables and calculates the target index used to evaluate the operating effect; then, it applies a small disturbance near the current optimization variable, compares the evaluation results before and after the disturbance, and obtains the optimization direction; next, it updates the optimization variable according to a preset step size, and maps the update result to a phase shift compensation amount, which is then output as a control command. The above steps are repeated cyclically. When conditions such as the upper limit of the number of iterations or the change in the evaluation index is less than a threshold are met, the update stops or the current output is maintained, thereby achieving online optimization while ensuring system stability.

[0070] Reference Figure 2 The specific implementation of this step includes the following:

[0071] 5.1) The controller collects and acquires the current operating status variables, including DC side voltage Vdc, grid side voltage Vg, leakage inductance current iLk, and equivalent voltage Veq. These state variables are used to calculate the objective function J.

[0072] 5.2) Set the initial value of the optimization variable ξ0 in the controller, set the iteration parameters, including step size η, disturbance amount δξ, termination threshold ε, and set the iteration count n=0;

[0073] 5.3) The controller calculates the objective function J based on the operating state variables and the current optimization variable ξ0, which is used to characterize the current operating state;

[0074] 5.4) The controller applies positive and negative disturbances ±δξ near the optimization variable ξ0, and calculates the positive and negative disturbance values ​​of the objective function respectively, to obtain the positive disturbance value J+ and the negative disturbance value J− of the objective function, so as to obtain the trend information of the objective function J near the optimization variable ξ0.

[0075] 5.5) The controller obtains the gradient estimate g based on the positive and negative disturbance values ​​of the objective function: g=(J+-J-)·ΔZVS,k / 2δξ, where g is used to characterize the trend of the objective function with respect to ξ;

[0076] 5.6) The controller iteratively updates the optimization variable ξ0 according to the preset step size η to obtain the optimization variable for the current round: ξn+1=ξn−ηg;

[0077] 5.7) Update the count to n = n + 1, and repeat steps 5.4) - 5.6) to determine whether the iteration update result of the current round satisfies the iteration termination condition:

[0078] If |g|≤ε, then the termination condition of the current iteration is satisfied, and the optimal optimization variable ξ*=ξn+1 is output;

[0079] If |g| > ε, then the termination condition for the current iteration is not met, and steps 5.4)-5.6) continue to be repeated.

[0080] Step 6: Based on the soft-switching decision signal ΔZVS,k and the optimal optimization variable ξ*, calculate the corrected inner phase shift ΔD1', outer phase shift ΔD2' and the corrected inner phase shift D1' and outer phase shift D2' that satisfy the online optimization of return power and soft-switching constraints.

[0081] 6.1) Based on the optimal optimization variable ξ* and the soft-switching decision signal ΔZVS,k, calculate the corrected internal phase shift ΔD1' that satisfies the online optimization of return power and soft-switching constraints:

[0082] ,

[0083] Where k is the proportional coefficient of the optimal optimization variable ξ*, and its value is set to 0.5;

[0084] 6.2) Based on the calculated ΔD1', D1, D2 and the soft-switching decision signal ΔZVS,k, calculate the corrected external phase shift ΔD2' that satisfies the online optimization of return power and soft-switching constraints:

[0085] ;

[0086] 6.3) Based on the calculated D1 and ΔD1', calculate the corrected internal phase shift D1' that satisfies the online optimization of return power and soft-switching constraints:

[0087] ;

[0088] 6.4) Based on the calculated D2 and ΔD2', calculate the corrected external phasor D2' that satisfies the online optimization of return power and soft-switching constraints:

[0089] .

[0090] Step 7: Output the switching transistor drive signal based on the corrected inner phase shift D1' and outer phase shift D2'.

[0091] Reference Figure 3 The implementation of this step includes the following:

[0092] 7.1) Input the corrected inner phase shift D1', outer phase shift D2', and switching frequency fs;

[0093] 7.2) Integrate the switching frequency fs and take the fractional part to generate a normalized reference clock pwm_clk(t)∈[0,1), so as to provide a uniform phase coordinate in each switching cycle;

[0094] 7.3) Receive the corrected inner phase shift D1' and outer phase shift D2', obtain the starting phase parameter origin and ending phase parameter end of each bridge arm switch, and normalize the phase parameters to the relative time within the same period;

[0095] 7.4) Perform an interval comparison on the above-mentioned reference clock pwm_clk(t), the starting phase parameter origin, and the ending phase parameter end;

[0096] The output is high when origin ≤ pwm_clk(t) ≤ end.

[0097] If the condition is not met, output a low level.

[0098] 7.5) Based on the high and low level signals obtained from the interval comparison, a square wave pulse switching transistor drive signal with a fixed duty cycle of 50% is generated. The start time of each pulse is uniquely determined by the corresponding key action time point, and its turn-off time is automatically obtained by delaying the start time by half a switching cycle. This signal is used to drive the switching transistor in the dual active bridge micro-inverter circuit to work.

[0099] It should be noted that the step numbers and claim numbers in this example are only applicable to the complete and clear description of the solution of the present invention, and their order is not limited.

[0100] Example 2: Online Co-optimization System Based on Return Power and Soft Switching Constraints.

[0101] Reference Figure 4 This example includes a feedforward control module 1, a soft-switching online determination module 2, a return power online optimization module 3, and a phase-shifting modulator module 4. The return power online optimization module 3 includes an input parameter acquisition submodule 31, an equivalent voltage and objective function calculation submodule 32, a disturbance injection submodule 33, a gradient estimation and optimization update submodule 34, and a modulation mapping and output submodule 35. The phase-shifting modulator module 4 includes a reference clock generation submodule 41, a phase parameter normalization submodule 42, and a pulse generation submodule 43.

[0102] The working principle of the entire system is as follows:

[0103] The feedforward control module 1 is used to generate the inner phase shift quantity D1 and the outer phase shift quantity D2 required for phase shift control based on the state variables such as photovoltaic side voltage Vpv, grid side voltage Vg, grid side phase θ, grid side current ig and grid side current reference value Im,ref, and input the D1 and D2 as the basic instructions for online optimization of back-current power to the online optimization module 3.

[0104] The soft-switching online determination module 2 is used to sample the arm current ileg,k(t) in real time at the time of commutation event k, calculate the soft-switching related indicators based on the actual current Ik, output the soft-switching determination signal ΔZVS,k to characterize the soft-switching state at the time of commutation event k, and input the soft-switching determination signal ΔZVS,k to the return power online optimization module 3.

[0105] The online optimization module 3 for return power is used to iteratively update the optimization variable ξ online based on the inner phase shift D1, the outer phase shift D2, and the soft-switching judgment signal ΔZVS,k, and in conjunction with the real-time state variables, to obtain the optimal optimization variable ξ*, and to calculate the corrected inner phase shift D1′ and outer phase shift D2′. The input parameter acquisition submodule 31 is used to acquire the DC-side voltage Vdc, the grid-side voltage Vg, the leakage inductance current iLk(t), the inner phase shift D1, and the outer phase shift D2, and outputs the acquired parameters to the equivalent voltage and objective function calculation submodule 32. This equivalent voltage and objective function calculation submodule 32 calculates the equivalent voltage Veq(t) and the objective function J based on the DC-side voltage Vdc, the grid-side voltage Vg, and the leakage inductance current iLk(t), and outputs the calculation results to the disturbance injection submodule 33. This disturbance injection submodule 33 applies positive and negative disturbances ±δξ to the optimization variable ξ to obtain the positive disturbance value J+ and the objective function. The negative perturbation value J- is obtained by applying positive and negative perturbations ±δξ to obtain the positive perturbation value J+ and the negative perturbation value J- of the objective function, and output to the gradient estimation and optimization submodule 34. The gradient estimation and optimization submodule 34 is used to calculate the gradient estimate g and the online iterative update of the optimization variable ξ based on the positive and negative perturbation values ​​of the objective function and the positive and negative perturbations ±δξ, and output the updated optimal optimization variable ξ* to the modulation mapping and output submodule 35. The modulation mapping and output submodule 35 is used to calculate the corrected inner phase shift D1' and outer phase shift D2' based on the updated optimal optimization variable ξ* and the soft switching judgment signal ΔZVS,k, and output the corrected D1' and D2' to the subsequent phase shift modulation module 4.

[0106] The phase-shift modulator module 4 is used to generate a square wave pulse with a fixed duty cycle of 50% for each switch transistor based on the corrected inner phase shift D1', outer phase shift D2', and switching frequency fs, to drive the switch transistor to work. The start time of each pulse is uniquely determined by the corresponding critical action point, and its turn-off time is automatically derived by delaying the start time by half a switching cycle. The reference clock generation submodule 41 is used to integrate the switching frequency fs and take its fractional part to generate a normalized reference clock pwm_clk(t)∈[0,1), providing a unified phase coordinate within each switching cycle, and outputting the normalized reference clock to the phase parameter normalization module. The phase parameter normalization submodule 42 and the pulse generation submodule 43 are used to receive the corrected inner phase shift D1' and outer phase shift D2', obtain the starting phase parameter origin and ending phase parameter end of each bridge arm switch, normalize the phase parameters to the relative time within the same period, and output the normalized result of the starting phase parameter origin and ending phase parameter end to the pulse generation submodule 43. The pulse generation submodule 43 performs interval comparison on the three parameters, and outputs a high level when origin≤pwm_clk(t)≤end, otherwise outputs a low level to generate the drive signal of the switching transistor.

[0107] It should be noted that the above functional modules can be implemented entirely or partially through software or hardware. When implemented in software, they can be implemented entirely or partially in the form of program instruction products. A program instruction product includes one or a set of program instructions. When the program instructions are loaded and executed on a computer, the described process or function is generated entirely or partially. The computer can be a general-purpose computer, a special-purpose computer, or other programmable device. Program instructions can be stored in a computer-readable and writable storage medium, or transferred from one computer's readable and writable storage medium to another.

[0108] The direct coupling or communication connections between the modules shown or discussed in this embodiment can be achieved through indirect coupling or communication connections via interfaces, devices, or modules. The various functional modules and sub-modules in this embodiment can dynamically reside within a single processing unit, or each module can exist physically independently, or two or more modules can dynamically reside within a single processing unit. When these dynamic components are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable and writable storage medium. This storage medium can be a memory, disk, or optical disc, etc.

[0109] The above description is merely two specific examples of the present invention and does not constitute any limitation on the present invention. Obviously, those skilled in the art, after understanding the content and principle of the present invention, may make various modifications and changes in form and detail without departing from the principle and structure of the present invention. For example, the switching frequency fs, photovoltaic side voltage Vpv, grid side voltage Vg, DC side voltage Vdc, the turns ratio n of the secondary winding to the primary winding of the transformer, the reference value Im,ref of the grid side current, the power weighting coefficient w1, the weighting coefficient w2 of the effective value of the inductor current, and the proportional coefficient k of the optimal optimization variable ξ∗ can all be modified according to requirements, except for the parameters set in this example. However, these modifications and changes based on the concept of the present invention are still within the scope of protection of the claims of the present invention.

Claims

1. An online collaborative optimization method based on return power and soft-switching constraints, characterized in that, include: S1) Sample the real-time state variables of the dual active bridge micro-inverter circuit, including: photovoltaic side voltage Vpv, DC side voltage Vdc, photovoltaic side current ipv, grid side voltage Vg, grid side phase θ, and bridge arm commutation current i at commutation event k. leg,k (t), leakage inductance current i Lk (t) and the bridge arm equivalent voltage Viq(t); S2) Based on sampled photovoltaic side voltage Vpv, photovoltaic side current ipv, and leakage inductance current i Lk Given the bridge arm equivalent voltage Veq(t), grid-side voltage Vg, and grid-side phase θ, calculate the inner phase shift D1, outer phase shift D2, and objective function J; S3) Based on the sampled bridge arm commutation current i leg,k (t), calculate the actual current Ik; S4) Based on the sign of Ik, output the soft-switching judgment signal at time k of the commutation event. Then, based on the soft-switching judgment signal, perform online iterative updates on the current value of the optimization variable ξ to obtain the optimal optimization variable ξ. * ; S5) Based on the updated optimal optimization variable ξ * The corrected inner phase shift D1' and outer phase shift D2' are calculated, and the drive signal of the switching transistor is output through the phase shift modulator to realize the online collaborative optimization method of return power and soft switching constraint.

2. The method according to claim 1, characterized in that, In step (S1), the real-time state variables of the dual active bridge micro-inverter circuit are sampled using an ideal voltage measurement unit, current measurement unit, and phase-locked loop unit provided by simulation software. The sampling process for each state variable is as follows: The photovoltaic side voltage Vpv, DC side voltage Vdc, grid side voltage Vg, and bridge arm equivalent voltage Veq(t) are obtained by parallel ideal voltage measurement unit. This unit is connected in parallel across the two ends of the signal being measured in the form of high input impedance. The measurement port is connected between the target node and the reference node, and outputs a continuous-time voltage signal V(t). Photovoltaic side current IPV, leakage inductance current i Lk (t) and the arm commutation current i at time k of commutation event. leg,k (t) is obtained by connecting an ideal current measurement unit in series. This unit is connected in series to the branch of the signal being measured with zero series impedance, i.e., zero equivalent voltage drop, and outputs a continuous-time current signal i(t). The grid-side phase θ is obtained by real-time processing of the sampled grid-side voltage Vg through a second-order generalized integrator phase-locked loop (SOGI-PLL) unit in software. This unit first conditions and discretizes the grid-side voltage Vg before inputting it into the SOGI-PLL to generate two mutually orthogonal vd and vq signals. The vq signals are then proportionally and integrally adjusted to obtain the frequency correction amount, which is added to the rated grid-side angular frequency w0 to obtain the estimated angular frequency. The phase is updated by integrating the estimated angular frequency, and continuous tracking of the grid-side phase is achieved through this closed-loop process.

3. The method according to claim 1, characterized in that, In step (S2), the formulas for calculating the inner phase shift D1, the outer phase shift D2, and the objective function J are as follows: ； ； ； Where max{} is the maximum value function, θ is the phase of the grid-side voltage, Vg is the grid-side voltage, n is the turns ratio of the transformer's secondary winding to the primary winding, Vdc is the DC-side voltage, Vpv is the photovoltaic-side voltage, fs is the switching frequency of the micro-inverter, Lk is the leakage inductance value of the transformer's secondary winding, Im,ref is the reference value of the grid-side current, sgn(Vg) is the sign function of the grid-side voltage, and i is the leakage inductance current. Lk (t), bridge arm equivalent voltage Veq(t), w1 is the weighting coefficient of return power and w2 is the weighting coefficient of the effective value of inductor current.

4. The method according to claim 1, characterized in that, The formula for calculating the actual current Ik in step (S3) is as follows: ， Where sk∈{+1,-1} represents the direction of current required at time k of this commutation event, i leg,k (t) represents the bridge arm commutation current at time k of commutation event.

5. The method according to claim 1, characterized in that, In step (S4), based on the sign of Ik, a soft-switching determination signal at time k of the commutation event is output. The value of the optimization variable ξ at the current time is then iteratively updated online based on the soft-switching determination signal to obtain the optimal optimization variable ξ. * Its implementation includes the following: S41) Based on the sign of the actual current Ik, output the soft-switching judgment signal ΔZVS,k: If Ik < 0, then the output soft-switching decision signal ΔZVS,k is 1. Otherwise, the output soft-switching decision signal ΔZVS,k is 0; S42) Determine the gradient estimate g based on the optimization variable ξ and the soft-switching decision signal ΔZVS,k: ； Where δ is the disturbance coefficient and J() is the objective function; S43) Based on the gradient estimate g, the optimization variable ξ is updated iteratively, and the iterative formula is as follows: ； Where η is the iteration step size for the optimization variable ξ; S44) Based on the idea of ​​iterative update, update the result ξ of the iteration. n+1 As the optimal optimization variable ξ * .

6. The method according to claim 1, characterized in that, The (S5) is based on the optimal optimization variable ξ * Given the soft-switching determination signal ΔZVS,k, calculate the corrected inner phase shift D1' and outer phase shift D2', with the following formulas: ， ； Where k is the optimization variable ξ * The proportional coefficients, D1 is the inner phase shift, D2 is the outer phase shift, and ΔZVS,k is the soft-switching decision signal.

7. The method according to claim 1, characterized in that, In step (S5), based on the corrected inner phase shift D1', outer phase shift D2', and switching frequency fs, the driving signal for the switching transistor is output through the phase shift modulator. This is achieved by: Based on the corrected inner phase shift quantity D1', outer phase shift quantity D2', and switching frequency fs, a square wave pulse with phase shift control is generated by a phase shift modulator to drive the switching transistors in the dual active bridge micro-inverter circuit.

8. An online collaborative optimization system based on return current power and soft-switching constraints, characterized in that, include: The feedforward control module is used to generate the inner phase shift quantity D1 and the outer phase shift quantity D2 required for phase shift control based on the measured photovoltaic side voltage Vpv, grid side voltage Vg, grid side phase θ, grid side current ig and the set grid side current reference value Im,ref. These serve as the basic instructions for subsequent soft switching determination and phase shift modulation. The soft-switching online determination module is used to sample the arm current i in real time at commutation event k. leg,k (t), calculate the soft-switching related indicators based on the actual current Ik, and output the soft-switching judgment signal; The online optimization module for return current power is used to iteratively update the optimization variable ξ online based on the soft-switching state output by the online soft-switching determination module, in order to obtain the optimal optimization variable ξ. * Calculate the corrected inner phase shift D1' and outer phase shift D2'; The phase-shift modulator module is used to generate a square wave pulse with a fixed duty cycle of 50% for each switch based on the inner phase shift D1', the outer phase shift D2' and the switching frequency fs. The start time of each pulse is uniquely determined by the corresponding critical action point, and its turn-off time is automatically derived by delaying the start time by half a switching cycle.

9. The system according to claim 8, characterized in that, The online optimization module for return power includes: The input parameter acquisition submodule is used to acquire the DC side voltage Vdc, the grid side voltage Vg, and the leakage inductance current i. Lk (t), inner phase shift amount D1 and outer phase shift amount D2; The equivalent voltage and objective function calculation submodule is used to calculate the equivalent voltage based on the DC-side voltage Vdc, grid-side voltage Vg, and leakage inductance current i. Lk Calculate the equivalent voltage Veq(t) and the objective function J; The perturbation injection submodule is used to apply positive and negative perturbations ±δξ to the optimization variable ξ to obtain the positive perturbation value J of the objective function. + and the negative perturbation value J of the objective function - ; The gradient estimation and optimization submodule is used to calculate the gradient estimate g and the online iterative update of the optimization variable ξ based on the positive and negative perturbation values ​​and the perturbation amount δξ of the objective function. The modulation mapping and output submodule is used to adjust the updated optimal optimization variable ξ. * The soft-switching determination signal ΔZVS,k is used to calculate the corrected inner phase shift D1' and outer phase shift D2'.

10. The system according to claim 8, characterized in that, The phase-shifting modulator module includes: The reference clock generation submodule is used to integrate the switching frequency fs and take the fractional part to generate a normalized reference clock pwm_clk(t)∈[0,1), so as to provide a uniform phase coordinate in each switching cycle. The phase parameter normalization submodule is used to receive the corrected inner phase shift D1' and outer phase shift D2', obtain the starting phase parameter origin and ending phase parameter end of each bridge arm switch, and normalize the phase parameters to the relative time within the same period; The pulse generation submodule is used to perform interval comparison of the reference clock pwm_clk(t), the start phase parameter origin, and the end phase parameter end; when origin≤pwm_clk(t)≤end, it outputs a high level, otherwise it outputs a low level to generate the drive signal for the switching transistor.

Images