Bidirectional DC / DC converter state monitoring method based on digital twinning

Abstract

The invention discloses a bidirectional DC / DC converter state monitoring method based on digital twinning. The method comprises the following steps: establishing a digital twin model based on a state space model; the inductive current and the output voltage of the bidirectional DC / DC converter in actual operation are collected; according to the inductive current, the output voltage, the inductance initial value, the capacitance initial value and the resistance initial value during actual operation, a digital twinborn model is utilized to obtain calculated values of the inductive current and the output voltage; comparing the actual value with the calculated value, and calculating a model error; continuously iteratively updating the inductance, capacitance and resistance of the bidirectional DC / DC converter according to the model error; the model error meets the preset error requirement; and comparing the inductor, the capacitor and the resistor which are iteratively updated for the last time with the inductors, the capacitors and the resistors under normal conditions and various fault conditions, so as to realize the detection of the state of each element. According to the scheme of the invention, fault detection of each element of the bidirectional DC / DC converter can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power industry, and in particular to a bidirectional DC / DC converter state monitoring method based on digital twin. Background Art

[0002] The UAV energy system provides power support for airborne equipment such as electric actuators, flight control, steering gear, and mission payloads. The reliable operation of the energy system is the basic guarantee for the normal operation of the UAV. Bidirectional DC / DC converters are commonly used in UAV energy systems, and their high failure rate poses a serious threat to the reliable operation of UAV energy systems.

[0003] Capacitor failure is the main factor leading to converter failure. If there is a problem with the capacitor on the input side of the module, the system will not be able to control the ripple voltage within the specified range. At the same time, as the energy storage part, it cannot provide effective power decoupling and power-off protection. If the filter capacitor on the output side of the module fails, it will affect the waveform quality of the output voltage, resulting in an increase in the ripple amplitude and a decrease in system efficiency. At the same time, if other components such as switching tubes and inductors fail, they will also greatly change the converter circuit topology, causing the converter parameter characteristics to degrade significantly, thus failing to meet normal working requirements and affecting the stable operation of the drone energy system. Therefore, accurate monitoring of the real-time working status of each component of the converter is crucial for the stable operation of the system. However, traditional monitoring methods face many challenges, such as the difficulty in obtaining data due to harsh environments, the complexity and nonlinear characteristics of machine learning algorithms, which make the model difficult to understand and interpret, and the amount of calculation is large. These problems affect the effect of real-time monitoring and prediction. Summary of the invention

[0004] In order to solve the technical problems of incomplete feature extraction, difficult model understanding and interpretation, and large amount of calculation in traditional bidirectional DC / DC converter monitoring methods, an embodiment of the present invention provides a bidirectional DC / DC converter state monitoring method based on digital twins.

[0005] The technical solution of the embodiment of the present invention is achieved as follows: An embodiment of the present invention provides a bidirectional DC / DC converter state monitoring method based on digital twins, the method comprising: A digital twin model based on a state space model is established; the digital twin model is: ; in, is the output voltage at the n+1th time step, is the output voltage at the nth time step, is the inductor current at the nth time step; , , are the coefficients of the state-space model obtained by using the fourth-order Runge-Kutta method, and are given by L.C and R Calculate and obtain; L is the inductor of the bidirectional DC / DC converter, C is the capacitance of the bidirectional DC / DC converter, R is the resistance of the bidirectional DC / DC converter; The inductor current and output voltage of the bidirectional DC / DC converter are collected during actual operation; and the collected inductor current and output voltage are divided according to time steps to obtain data pairs consisting of a plurality of actual values ​​of the output voltage in a previous time step, an actual value of the inductor current in a previous time step, and an actual value of the output voltage in a next time step; For each data pair, the actual value of the output voltage at the previous time step and the actual value of the inductor current at the previous time step are input into the digital twin model, and the initial value of the inductance, the initial value of the capacitance and the initial value of the resistance of the bidirectional DC / DC converter generated by the genetic algorithm are input into the digital twin model to calculate the output voltage calculation value at the next time step; Calculate the model error based on the actual output voltage value of the next time step of the plurality of data pairs and the calculated output voltage value of the next time step and using a preset error model; Determine whether the model error meets the preset error requirement; if it is determined that the model error does not meet the preset error requirement, use the taboo search algorithm to update the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter, and recalculate the output voltage calculated value of each data pair in the next time step based on the updated inductance value, capacitance value and resistance value of the bidirectional DC / DC converter; recalculate the model error according to the actual value of the output voltage in the next time step of multiple data pairs and the output voltage calculated value in the next time step, using the preset error model; determine whether the recalculated model error meets the preset error requirement; if it is determined that the recalculated model error does not meet the preset error requirement, continue to use the taboo search algorithm to update the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter until it is determined that the model error meets the preset error requirement; Obtain the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter updated last time; continue to compare the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter updated last time with the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter under normal conditions and the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter under fault conditions, respectively, to determine the fault type of the bidirectional DC / DC converter.

[0006] In one embodiment, a digital twin model based on a state space model is established, including: According to the two working modes of the bidirectional DC / DC converter Buck and Boost, a state space model is constructed by using the state space averaging method; and a differential equation expression of the state space model is established; The differential equation expression of the state space model is linearized by using a fourth-order Runge-Kutta pair differential equation to obtain an iterative expression between the inductor current at the n+1th step and the inductor current at the nth step, and an iterative expression between the output voltage at the n+1th step and the output voltage at the nth step; The iterative expression between the inductor current of the n+1th step and the inductor current of the nth step, as well as the iterative expression between the output voltage of the n+1th step and the output voltage of the nth step are substituted into the differential equation expression of the state-space model to obtain a digital twin model based on the state-space model.

[0007] In one embodiment, the state space model includes: ; ; Among them, V1 represents the voltage stabilization value of the high-voltage side, Represents the transient value of the high-voltage side voltage, represents the transient value of the low-voltage side voltage, C1 and C2 represent the capacitance of the high-voltage side and the low-voltage side respectively. L is the inductance, D is the duty cycle of the converter, i L is the inductor current, R1 and R2 are the loads in Buck and Boost modes respectively. Indicates the instantaneous value of the inductor current.

[0008] In one embodiment, the differential equation expression of the state-space model includes: ; Among them, i L is the inductor current, u o is the output voltage, represents the inductor voltage, Indicates the output voltage.

[0009] In one embodiment, the iterative expression between the inductor current of the n+1th step and the inductor current of the nth step, and the iterative expression between the output voltage of the n+1th step and the output voltage of the nth step include: ; in, is the inductor current at step n+1, is the inductor current at step n, is the output voltage of the n+1th step, is the output voltage of the nth step, k1~k4 and z1~z4 are used to calculate the average rate of change between the nth step and the n+1th step, the formula is as follows: ; h is the computational step size between the nth and n+1th time intervals.

[0010] In one embodiment, the preset error model includes: ; in, is the model error, N is the total number of collected data; I L，i and U o，i is the calculated value of the inductor current and the calculated value of the output voltage in the next time step; l Lm，i and U om，i is the actual value of the inductor current and the actual value of the output voltage in the next time step.

[0011] This embodiment has the following beneficial effects: This embodiment first builds a digital twin model based on the state space model based on the digital twin technology to accurately simulate the working status of each component of the bidirectional DC / DC converter; then, based on the digital twin model, combined with the genetic algorithm and the taboo search algorithm, the real-time operating data of each component of the real bidirectional DC / DC converter is compared with the data under normal conditions and various fault conditions to monitor the fault conditions of each component of the bidirectional DC / DC converter in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a flow chart of a method for monitoring the state of a bidirectional DC / DC converter based on digital twins according to an embodiment of the present invention; Figure 2 Schematic diagram of a bidirectional DC / DC converter according to an embodiment of the present invention; Figure 3 Schematic diagram of Buck mode of bidirectional DC / DC converter according to an embodiment of the present invention; Figure 4 Schematic diagram of a bidirectional DC / DC converter Boost mode according to an embodiment of the present invention; Figure 5 A schematic diagram of a genetic algorithm flow chart of an embodiment of the present invention; Figure 6 Schematic diagram of the process of the taboo search algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0013] The present invention will be described in further detail below with reference to the accompanying drawings and embodiments.

[0014] The embodiment of the present invention provides a bidirectional DC / DC converter state monitoring method based on digital twin, such as Figure 1 As shown, the method includes: Step 101: Establish a digital twin model based on a state space model; the digital twin model is: ; in, is the output voltage at the n+1th time step, is the output voltage at the nth time step, is the inductor current at the nth time step; L is the inductor of the bidirectional DC / DC converter, C is the capacitance of the bidirectional DC / DC converter, R is the resistance of the bidirectional DC / DC converter; Step 102: collecting the inductor current and the output voltage of the bidirectional DC / DC converter when it is actually running; and dividing the collected inductor current and the output voltage according to time steps to obtain a plurality of data pairs consisting of an actual value of the output voltage in a previous time step, an actual value of the inductor current in a previous time step, and an actual value of the output voltage in a next time step; Step 103: For each data pair, the actual value of the output voltage at the previous time step and the actual value of the inductor current at the previous time step are input into the digital twin model, and the initial value of the inductance, the initial value of the capacitance and the initial value of the resistance of the bidirectional DC / DC converter generated by the genetic algorithm are input into the digital twin model to calculate the output voltage calculated value at the next time step; Step 104: Calculate the model error based on the actual output voltage value of the next time step of the plurality of data pairs and the calculated output voltage value of the next time step and using a preset error model; Step 105: judging whether the model error meets the preset error requirement; if it is judged that the model error does not meet the preset error requirement, using the taboo search algorithm to update the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter, and recalculating the output voltage calculated value of each data pair in the next time step based on the updated inductance value, capacitance value and resistance value of the bidirectional DC / DC converter; recalculating the model error according to the actual output voltage value of the next time step of multiple data pairs and the output voltage calculated value of the next time step, using the preset error model; judging whether the recalculated model error meets the preset error requirement; if it is judged that the recalculated model error does not meet the preset error requirement, continuing to use the taboo search algorithm to update the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter until it is judged that the model error meets the preset error requirement; Step 106: Obtain the inductance value, capacitance value, and resistance value of the bidirectional DC / DC converter that were last updated; continue to compare the inductance value, capacitance value, and resistance value of the bidirectional DC / DC converter that were last updated with the inductance value, capacitance value, and resistance value of the bidirectional DC / DC converter under normal conditions and the inductance value, capacitance value, and resistance value of the bidirectional DC / DC converter under fault conditions, respectively, to determine the fault type of the bidirectional DC / DC converter.

[0015] This embodiment first builds a digital twin model based on the state space model based on the digital twin technology to accurately simulate the working status of each component of the bidirectional DC / DC converter; then, based on the digital twin model, combined with the genetic algorithm and the taboo search algorithm, the real-time operating data of each component of the real bidirectional DC / DC converter is compared with the data under normal conditions and various fault conditions to monitor the fault conditions of each component of the bidirectional DC / DC converter in real time.

[0016] Next, this embodiment will introduce the construction process of the digital twin model based on the state space model.

[0017] Step 1: Modeling the Bidirectional DC / DC Converter Topology Specifically, see Figure 2 , Figure 2 The schematic diagram of the bidirectional DC / DC converter is shown in Figure 2. The small signal model of the bidirectional DC / DC converter is first derived using the state space averaging method, thereby establishing a digital twin model based on the state space. Since the bidirectional DC / DC converter has two working modes, Buck and Boost, they are derived separately: (1) Buck mode: ; Buck mode small signal model schematic diagram is as follows Figure 3 shown.

[0018] (2) Boost mode: ; The schematic diagram of the boost mode small signal model is as follows: Figure 4 shown.

[0019] Where V1 represents the high voltage side voltage, V2 represents the low voltage side voltage, C1 and C2 represent the capacitance of the high voltage side and the low voltage side respectively, L is the inductance, D is the duty cycle of the converter, i L is the inductor current, R1 and R2 are the loads in Buck and Boost modes respectively.

[0020] Step 2: Solve the model parameters using the fourth-order Runge-Kutta formula The Runge-Kutta method is a numerical method commonly used to numerically solve ordinary differential equations (ODEs). The Runge-Kutta method discretizes the differential equations and converts the continuous solution problem into a discrete step problem. This method divides time or independent variables into multiple small steps, also called time steps or step points, and then calculates the change of the variable in each step. These changes are weighted averaged according to certain weights to obtain an approximate solution for the next time step. This method has the advantages of high accuracy, strong stability and wide application range.

[0021] The state space model is simplified into the following differential equations based on Buck and Boost modes respectively: ; Among them, i L is the inductor current, u o is the output voltage.

[0022] The fourth-order Runge-Kutta method is used to linearize the differential equation to calculate the inductor current at the n+1th step of the iteration. and output voltage , and it is believed that the error of this method can be ignored: ; Among them, k1~k4 and z1~z4 are used to calculate the average rate of change between the nth step and the n+1th step. The formula is as follows: ; h is the calculation step between the nth and n+1th time intervals. By substituting it into the state space average model, the final digital twin model based on the state space model can be obtained: ; in, is the output voltage at the n+1th time step, is the output voltage at the nth time step, is the inductor current at the nth time step; L is the inductor of the bidirectional DC / DC converter, C is the capacitance of the bidirectional DC / DC converter, R is the resistance of the bidirectional DC / DC converter.

[0023] Since the formulas are constantly nested, the calculation formulas for a, b, and c are relatively complex. In the formula, the parameters a, b, and c are actually composed of a parameter set (L, C, R, Vin, Uo, IL, D). Among them, L, R, and C are unknown parameters; D and Vin are known parameters, D is a given switch device parameter; Vin is the input voltage measured by the sensor.

[0024] According to the above digital twin model, the output voltage at the n+1th step can be calculated using the inductor current and output voltage at the nth step.

[0025] The above is the construction process of the digital twin model based on the state space model. Based on the above digital twin model, the following will introduce how the improved taboo search algorithm can realize the fault monitoring of each component of the bidirectional DC / DC converter.

[0026] The inductance, capacitance and resistance of each component of a bidirectional DC / DC converter will change significantly under normal or fault conditions. However, during the operation of the bidirectional DC / DC converter, it is impossible to detect the specific inductance, capacitance and resistance. Only the actual values ​​of the output voltage and inductor current can be detected indirectly.

[0027] Therefore, this embodiment first uses a genetic algorithm to generate an initial value of an inductor, capacitor, and resistor (i.e., an initial value of the inductor, an initial value of the capacitor, and an initial value of the resistor), and then inputs the initial value of the inductor, the initial value of the capacitor, and the initial value of the resistor, as well as the detected output voltage and inductor current into the digital twin model to calculate the output voltage and inductor current of the next time step; the calculated output voltage and inductor current are compared with the output voltage and inductor current of the next time step collected when the bidirectional DC / DC converter is actually running (i.e., the calculated value is compared with the true value), and the values ​​of the inductor, capacitor, and resistor are iterated according to the comparison result by using a taboo search algorithm, so as to obtain the values ​​of the inductor, capacitor, and resistor when the calculated value is closest to the true value, that is, the digital twin model is used to effectively complete the simulation of the actual operation of the bidirectional DC / DC converter, and the estimated values ​​of the inductor, capacitor, and resistor when the bidirectional DC / DC converter is actually running are obtained. By comparing the estimated values ​​of the inductor, capacitor, and resistor with the values ​​of the inductor, capacitor, and resistor in normal conditions or fault conditions, the fault type can be determined.

[0028] This embodiment uses a simulation model to simulate the actual operation of a bidirectional DC / DC converter. First, the values ​​of various parameters such as input voltage, resistance R, inductance L, capacitance C, and converter duty cycle D are set in the bidirectional DC / DC circuit simulation model, and the input voltage is collected and transmitted to the digital twin model, and the collected inductor current and capacitor voltage in the simulation circuit are transmitted to the taboo search algorithm. The corresponding inductor current and capacitor voltage calculated by the digital twin model are also transmitted to the taboo search algorithm. The taboo search algorithm randomly generates an initial solution, substitutes the data of each solution into the preset error model, and obtains the optimal set of solutions for the preset error model, which is the result of identifying the parameters inductor L, resistance R, and capacitance C.

[0029] Here, this embodiment constructs the following preset error model, that is, using variance to calibrate the error between the iteration parameter and the actual model: ; in, is the model error, N is the total number of collected data; I L，i and U o，i is the calculated value of the inductor current and the calculated value of the output voltage in the next time step; l Lm，i and U om，i is the actual value of the inductor current and the actual value of the output voltage in the next time step.

[0030] Here, in this embodiment, I L，i and U o，i The inductor current and capacitor voltage are calculated by the digital twin model based on the data; Lm，i and U om，i Inductor current and capacitor voltage measured for a bidirectional DC / DC circuit simulation model.

[0031] In addition, in order to reduce the amount of calculation, this embodiment improves the taboo search algorithm. The unimproved taboo search algorithm randomly generates inductance L, resistance R, and capacitance C and inputs them into the digital twin model. This embodiment introduces a genetic algorithm, which uses the genetic algorithm to obtain the initial values ​​of the inductance L, resistance R, and capacitance C, ensuring that the initial values ​​are closer. The initial values ​​of the inductance L, resistance R, and capacitance C are input into the digital twin model, and the inductance current and capacitance voltage are calculated. The preset error model is used to calculate and determine whether the variance is within the error. If not, the taboo search algorithm continues to iterate the values ​​of the inductance L, resistance R, and capacitance C until the error requirement is met. The output, then the output values ​​of the inductance L, resistance R, and capacitance C are compared with the values ​​of the inductance L, resistance R, and capacitance C in normal or fault conditions to determine the type of fault.

[0032] See also Figure 5 and Figure 6 , the process of using the improved taboo search algorithm to monitor the status of each component is: The taboo search algorithm is highly dependent on the initial solution, that is, when the initial solution is good, the optimal solution can be found quickly. When the initial solution is not good, the speed of the taboo search is directly restricted. In order to improve the method, the genetic algorithm is introduced. The genetic algorithm has the advantages that the optimization results are independent of the initial conditions and the algorithm is independent of the solution domain, but the convergence speed is slow, so it is only used to obtain the initial solution of the taboo search algorithm. The steps for implementing the genetic algorithm programming are as follows: (1) Coding. Binary coding is used to encode the estimated parameters. The circuit model has four unknown parameters. Each chromosome individual is set to have four gene segments, and each parameter corresponds to a gene segment with eight gene bits. The chromosome consists of 24 gene bits in total. The meaning of each bit is shown in Table 1. Assume that the ranges of capacitance C, inductance L, and resistance RL are 470~130μF, 20~30μH, and 4~0.5Ω, respectively. The range of each parameter is used as its encoding and decoding interval.

[0033] Table 1

[0034] (2) Selection operator. The “roulette wheel” selection method is used, and the probability of each individual entering the next generation is the ratio of its fitness value to the sum of the fitness values ​​of the individuals in the entire population.

[0035] (3) Fitness function and fitness evaluation. Select equation (7) as the fitness function. Select the individual with the smallest sum of squared errors in each generation as the optimal individual of this generation, and compare the optimal individual of this generation with the overall optimal individual in the evolutionary process. The individual with the smallest sum of squared errors is the overall optimal individual.

[0036] (4) Crossover operation. In each generation of individuals, if the generated random number is less than the set crossover probability Pc = 0.6, two chromosomes are randomly selected, and the crossover point is randomly selected. Genes are exchanged at the crossover point to achieve crossover of the parent generation.

[0037] (5) Mutation operation. The mutation process is to set the initial mutation probability Pm = 0.1 and gradually reduce the mutation probability as the number of evolutionary generations increases.

[0038] (6) Termination conditions. The termination generation and fitness error are used as the termination conditions of the evolution process. The termination generation is set to 100 generations and the fitness error is set to 0.135.

[0039] The process of genetic algorithm is shown in Figure 5 .

[0040] In the previously established digital twin mathematical model of the bidirectional DC / DC topology circuit and its numerical solution method, due to the presence of three highly nonlinear unknown variables, it is difficult for traditional algorithms to achieve accurate parameter identification. As a classic optimization method, the taboo search algorithm has a strong local search capability and can avoid falling into the local optimal solution, thereby finding the global optimal solution. The application process of this algorithm in this article is as follows Figure 6 shown.

[0041] By adjusting the size and update strategy of the taboo table, the taboo search algorithm can flexibly explore the solution space and effectively balance the breadth and depth of exploration, thereby improving the efficiency and quality of the solution. The use of the taboo table is crucial to the stability and convergence of the algorithm, which helps to overcome the problem of local optimality and find a better solution. Substituting the data of each solution into the objective function, the optimal solution is the result of the parameters L, R, and C that need to be identified. Finally, the parameters estimated by the digital twin model are compared with the R, L, and C values ​​obtained under normal conditions and various fault conditions through circuit simulation, realizing the detection of the status of each component.

[0042] This embodiment first establishes the state space equation of the bidirectional DC / DC topology circuit, and constructs the corresponding digital twin model using the Runge-Kutta numerical solution method. Through the verification of the simulation model, the input and output data prove the accuracy of the digital twin model, ensuring the real-time monitoring stability of the UAV power supply. The taboo search algorithm is innovatively introduced in combination with the digital twin technology for parameter identification of the bidirectional DC / DC topology circuit. Finally, the parameters estimated by the digital twin model are compared with the R, L, and C values ​​obtained by circuit simulation under normal conditions and various fault conditions, realizing the detection of the status of each component.

[0043] Compared with traditional methods, modeling bidirectional DC / DC converters through digital twin technology can effectively track the output of the circuit and ensure the accuracy and stability of real-time monitoring of the UAV power supply. Using digital twin technology, real-time monitoring of the UAV power supply topology circuit can be achieved, and the operating parameters of each component can be identified to evaluate the performance of the power supply, study indicators such as failure rate, average life and failure mode, and finally derive the overall reliability level of the power supply, and optimize the design and production process accordingly.

[0044] The taboo search algorithm effectively avoids the problem of local optimal solution through the characteristics of taboo table and neighborhood solution, ensures the rapidity and accuracy of parameter identification, and the error is within an acceptable range. With the iteration of the algorithm, the identification accuracy is gradually improved.

[0045] The UAV power system has high requirements for stability and reliability, which can be met through accurate modeling and parameter identification. When designing a UAV, it is necessary to pay attention to the reliability of each system and perform corresponding model construction and parameter identification. The use of digital twin technology based on taboo search can effectively establish a power system model and realize real-time monitoring and parameter identification, thereby improving the reliability of the power supply and ensuring the smooth completion of the mission.

[0046] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0047] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A bidirectional DC / DC converter state monitoring method based on digital twin, characterized in that: The method comprises: A digital twin model based on a state space model is established; the digital twin model is: ； in, is the output voltage at the n+1th time step, is the output voltage at the nth time step, is the inductor current at the nth time step; , , are the coefficients of the state-space model obtained by using the fourth-order Runge-Kutta method, and are given by L.C and R Calculate and obtain; L is the inductor of the bidirectional DC / DC converter, C is the capacitance of the bidirectional DC / DC converter, R is the resistance of the bidirectional DC / DC converter; The inductor current and output voltage of the bidirectional DC / DC converter are collected during actual operation; and the collected inductor current and output voltage are divided according to time steps to obtain data pairs consisting of a plurality of actual values ​​of the output voltage in a previous time step, an actual value of the inductor current in a previous time step, and an actual value of the output voltage in a next time step; For each data pair, the actual value of the output voltage at the previous time step and the actual value of the inductor current at the previous time step are input into the digital twin model, and the initial value of the inductance, the initial value of the capacitance and the initial value of the resistance of the bidirectional DC / DC converter generated by the genetic algorithm are input into the digital twin model to calculate the output voltage calculation value at the next time step; Calculate the model error based on the actual output voltage value of the next time step of the plurality of data pairs and the calculated output voltage value of the next time step and using a preset error model; Determine whether the model error meets the preset error requirement; if it is determined that the model error does not meet the preset error requirement, use the taboo search algorithm to update the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter, and recalculate the output voltage calculated value of each data pair in the next time step based on the updated inductance value, capacitance value and resistance value of the bidirectional DC / DC converter; recalculate the model error according to the actual value of the output voltage in the next time step of multiple data pairs and the output voltage calculated value in the next time step, using the preset error model; determine whether the recalculated model error meets the preset error requirement; if it is determined that the recalculated model error does not meet the preset error requirement, continue to use the taboo search algorithm to update the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter until it is determined that the model error meets the preset error requirement; Obtain the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter updated last time; continue to compare the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter updated last time with the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter under normal conditions and the inductance value, capacitance value and resistance value of the bidirectional DC / DC converter under fault conditions, respectively, to determine the fault type of the bidirectional DC / DC converter.

2. The bidirectional DC / DC converter state monitoring method based on digital twin according to claim 1 is characterized in that: Establish a digital twin model based on the state space model, including: According to the two working modes of the bidirectional DC / DC converter Buck and Boost, a state space model is constructed by using the state space averaging method; and a differential equation expression of the state space model is established; The differential equation expression of the state space model is linearized by using a fourth-order Runge-Kutta pair differential equation to obtain an iterative expression between the inductor current at the n+1th step and the inductor current at the nth step, and an iterative expression between the output voltage at the n+1th step and the output voltage at the nth step; The iterative expression between the inductor current of the n+1th step and the inductor current of the nth step, as well as the iterative expression between the output voltage of the n+1th step and the output voltage of the nth step are substituted into the differential equation expression of the state-space model to obtain a digital twin model based on the state-space model.

3. The bidirectional DC / DC converter state monitoring method based on digital twin according to claim 2 is characterized in that: The state space model comprises: ； ； Among them, V1 represents the voltage regulation value of the high-voltage side, Represents the transient value of the high-voltage side voltage, represents the transient value of the low-voltage side voltage, C1 and C2 represent the capacitance of the high-voltage side and the low-voltage side respectively. L is the inductance, D is the duty cycle of the converter, i L is the inductor current, R1 and R2 are the loads in Buck and Boost modes respectively. Indicates the instantaneous value of the inductor current.

4. The bidirectional DC / DC converter state monitoring method based on digital twin according to claim 2 is characterized in that: The differential equation expression of the state space model includes: ； Among them, i L is the inductor current, u o is the output voltage, represents the inductor voltage, Indicates the output voltage.

5. The method for monitoring the state of a bidirectional DC / DC converter based on digital twin according to claim 2, characterized in that: The iterative expression between the inductor current of the n+1th step and the inductor current of the nth step, and the iterative expression between the output voltage of the n+1th step and the output voltage of the nth step, include: ； in, is the inductor current at step n+1, is the inductor current at step n, is the output voltage of the n+1th step, is the output voltage of the nth step, k1~k4 and z1~z4 are used to calculate the average rate of change between the nth step and the n+1th step, the formula is as follows: ； h is the computational step size between the nth and n+1th time intervals.

6. The method for monitoring the state of a bidirectional DC / DC converter based on digital twin according to claim 1, characterized in that: The preset error model includes: ； in, is the model error, N is the total number of collected data; I L，i and U o，i is the calculated value of the inductor current and the calculated value of the output voltage in the next time step; l Lm，i and U om，i is the actual value of the inductor current and the actual value of the output voltage in the next time step.

Images