CN-121979040-A - Efficient parameter identification method for double-active-bridge converter based on sparse state acquisition and fast state deduction

CN121979040ACN 121979040 ACN121979040 ACN 121979040ACN-121979040-A

Abstract

The invention relates to the technical field of dual-active bridge converter parameter identification, in particular to a dual-active bridge converter efficient parameter identification method based on sparse state acquisition and rapid state deduction, which comprises the steps of S101 defining model input, acquiring a system state of a DAB converter when modulated waves meet carrier waves each time under MCU conventional sampling configuration, S102, dynamic modeling of the DAB converter, S103, time step updating of the system state, discretization processing to obtain a system state evolution relation, S104, population initialization, S105, population target fitness evaluation, S106, genetic evolution, generation of a child population with a scale of M including pairing selection, crossing and mutation, S107 terminating condition judgment and optimal solution output, returning to S105 for continuous loop iteration when the fitness of an individual with optimal target performance tends to converge, and after the algorithm converges, the parameter with optimal individual performance on a main target is the final parameter identification result.

Inventors

Xiang Yangxiao
YUAN WEITING
XIE ZHUOJUN
DU XIONG
SUN PENGJU
LUO QUANMING

Assignees

重庆大学

Dates

Publication Date: 20260505
Application Date: 20260120

Claims (10)

1. The efficient parameter identification method of the double-active-bridge converter based on sparse state acquisition and fast state deduction is characterized by comprising the following steps of: s101, defining model input, under the MCU normal sampling configuration, acquiring system state when each switch of the DAB converter is started; s102, dynamically modeling the DAB converter, and constructing a continuous time state space model of the DAB converter; S103, performing discretization processing on the model by using a Dragon-Gregory tower method to obtain a system state evolution relation; s104, initializing a population, wherein the initial population consists of 2*M individuals, and each individual is a group of complete model parameter vectors to be identified; s105, evaluating population target fitness, constructing a single target fitness function, wherein the function is calculated based on the square sum of errors between the model calculated value obtained by deduction in the step S103 and the actually measured sampling value obtained in the step S101; S106, genetic evolution, including pairing selection, crossing and mutation, generating a offspring population with the scale M; s107, judging the termination condition and outputting the optimal solution, stopping the algorithm when the fitness of the individual with optimal performance at the target tends to converge, and returning to S105 to continue the loop iteration if the fitness of the individual with optimal performance at the target does not converge; and S108, extracting parameters, and after the algorithm converges, selecting an individual with the optimal performance on the main target, wherein the parameters are the final parameter identification result.
2. The method according to claim 1, wherein in step S101, the collected system state parameters include an inductor current i L , a secondary side output voltage V 2 , a switching sequence signal S 1 ,S 2 , a time point t of each switching state, and a time interval Δt.
3. The method for efficient parameter identification of a dual active bridge converter based on sparse state acquisition and fast state deduction according to claim 2, wherein in step S102, the continuous time state space model of the DAB converter is expressed as: The inductance current i L and the secondary side output voltage V 2 are system state variables of the DAB converter, parameters to be identified are an inductance L, a secondary side supporting capacitor C, a supporting capacitor parasitic resistance R C and a load resistance R, V 1 is a direct current power supply, and K is a transformer transformation ratio.
4. The method for identifying the efficient parameters of the dual-active-bridge converter based on sparse state collection and fast state deduction according to claim 3, wherein the specific steps of S103 are as follows: first, at each time step The q-stage intermediate slopes of i L and V 2 were calculated in: Next, the system state value at the next time t n+1 is calculated using the calculated q-group slope: Wherein the value of { a 11 ,…,a qq }, {b 1 ,…,b q }, {c 1 ,…,c q } refers to the RK table.
5. The efficient parameter identification method of a dual-active-bridge converter based on sparse state collection and fast state deduction according to claim 4, wherein in step S104, the model parameter vector to be identified is p= [ L, R c , C, R ], and the individuals in the initial population adopt a method of uniformly and randomly sampling in a logarithmic coordinate system.
6. The method for identifying the efficient parameters of the dual-active-bridge converter based on sparse state acquisition and fast state deduction according to claim 5, wherein the method for uniformly and randomly sampling the individuals in the initial population under a logarithmic coordinate system is as follows: Firstly, setting physical search boundaries of parameters according to design specifications of a DAB converter, wherein the physical search boundaries comprise an upper bound UB and a lower bound LB, then, generating a uniformly distributed random number in a logarithmic interval [ log 10 (LB j ),log 10 (UB j ] of each parameter p j to be generated, and finally, mapping the random number back to an original parameter value through an anti-logarithmic operation.
7. The method for efficient parameter identification of a dual active bridge converter based on sparse state acquisition and fast state deduction according to claim 6, wherein in step S105, the single target fitness function is: Wherein { is as follows , The { is the model calculation value deduced in step S103 , The } is the actual measurement sampling value obtained in step S101, N is the number of switching cycles, 4n+1 is the total number of sampling points, w is a weight coefficient of the difference of the magnitude order of the constant voltage and the current, and the smaller the fitness value J (P i ) is, the more excellent the individual is.
8. The method for identifying efficient parameters of dual active bridge converters based on sparse state acquisition and fast state deduction according to claim 7, wherein in step S106, the method for pairing selection is as follows: Randomly and repeatedly extracting two individuals from a parent population with the scale of M, comparing the merits of the two individuals, judging that the fitness value J (Pi) is small as a winning individual, putting the winning individual into a pairing pool, and repeating the pairing pool with the scale of M for M times; the parent population is an initial population or a child population generated in the last iteration.
9. The method for identifying efficient parameters of a dual active bridge converter based on sparse state collection and fast state deduction according to claim 8, wherein in step S106, the method of interleaving is to simulate binary interleaving for each pair of parent individuals P 1 =[p 1,1 ,...,p 1,j and P 2 =[p 2,1 ,...,p 2,j in a pairing pool; the method of simulating binary interleaving operation is as follows: Generating a random number u between [0,1 ]; according to a preset distribution index eta c , a spreading factor beta is calculated according to the following formula: (7) For each parameter p j in the parent vector, two new child parameters c 1,j and c 2,j are generated as follows: (8) (9)。
10. The method for identifying parameters of a dual-active bridge converter based on sparse state collection and fast state deduction according to claim 9, wherein in step S106, the method of interleaving operation is to use a polynomial mutation operator matched with a simulated binary interleaving operation, determine whether to mutate each sub-generation individual C generated after interleaving with a smaller mutation probability p m , and if so, independently operate each parameter C j in the individual, and the operation steps are as follows: Generating a random number r between [0,1 ]; Calculating a disturbance delta according to a preset variation distribution index eta m and the following formula; (10) Generating a mutated new parameter c ' j according to the following formula: (11) Here, UB j and LB j are the upper and lower search bounds of the parameter p j , and if c ' j crosses the bounds, it is set as the boundary value.

Description

Efficient parameter identification method for double-active-bridge converter based on sparse state acquisition and fast state deduction Technical Field The invention relates to the technical field of parameter identification of double-active-bridge converters, in particular to a method for identifying efficient parameters of a double-active-bridge converter based on sparse state acquisition and fast state deduction. Background The dual-active bridge (Dual Active Bridge, DAB) direct-current converter has become a core power conversion device of key application occasions such as an electric vehicle-mounted charger, a large-scale energy storage system, a solid-state transformer, a direct-current micro-grid and the like by virtue of the advantages of electric isolation, bidirectional power flow capability, high power density, easiness in realizing soft switching and the like. Under the long-term high-frequency and high-voltage complex working condition operation, the key physical parameters of the converter (such as a phase-shifting inductance value, leakage inductance of a high-frequency transformer, an output filter capacitance value, equivalent series resistance ESR (equivalent series resistance) of the high-frequency and high-voltage transformer, secondary side load resistance and the like) can generate remarkable dynamic drift under the comprehensive influence of the factors of the thermal effect of the magnetic element, dielectric aging, ambient temperature drift and the like. The accuracy of the parameters is not only the premise of realizing high-performance control strategies such as Model Predictive Control (MPC), but also the basis for carrying out the health status monitoring (SOH) and fault early warning of the converter. Therefore, the method builds a high-fidelity system model and realizes the online accurate identification of multiple parameters, and has extremely important significance for guaranteeing the safe and efficient operation of the DAB converter. However, the existing parameter identification technology still faces serious challenges in data acquisition and calculation methods. In the aspect of data acquisition, the DAB converter is used as a high-order nonlinear system, and the voltage and current waveform of the DAB converter contains rich high-frequency harmonic components. To obtain enough information to invert the system parameters, conventional methods typically rely on high sampling rate analog to digital converters (ADCs) or dedicated high frequency observation devices to capture the complete waveform details. However, in an industrial-level embedded control system (such as a DSP or MCU-based controller), the sampling frequency is often difficult to meet the requirements of high-frequency reconstruction due to hardware cost and chip processing power. In terms of calculation algorithm, aiming at the problem of multi-parameter identification, the existing solutions are mainly divided into two types of analytic deduction method and intelligent optimization algorithm, but limitations exist in the two types of analytic deduction method and intelligent optimization algorithm. Although the analysis method has high calculation speed, the analysis method excessively depends on an idealized mathematical model, has poor robustness to non-ideal factors (such as parasitic parameters), and the intelligent evolution algorithm represented by a Genetic Algorithm (GA) and a Particle Swarm Optimization (PSO) has global optimizing capability and can solve the nonlinear problem, but has larger calculation load. If the forward deduction is carried out by simply utilizing a numerical integration method, the dynamic process can be accurately simulated, but an automatic optimizing mechanism is lacked, and if the forward deduction is carried out by simply relying on an evolutionary algorithm, the forward deduction is easy to fall into local optimization. In summary, how to mine the deep physical features of the system under the limited condition of extremely sparse data is a key bottleneck for realizing low-cost and high-precision parameter identification of the DAB converter. The existing identification method does not find a balance point of cost and precision, speed and precision, and is difficult to popularize in a large scale in industrial products sensitive to cost. Therefore, how to realize the rapid and accurate identification of multiple parameters of the DAB converter by using sparse sampling data on the premise of not changing the conventional sampling configuration and not introducing additional high-frequency detection hardware is a technical problem to be overcome currently, and has great engineering application value and industrialization prospect. Disclosure of Invention Aiming at the problems that the existing dual-active-bridge converter parameter identification method relies on high-bandwidth sampling equipment to acquire dense data and is difficult to consider the calculation speed