Energy storage converter low voltage ride through control parameter identification method and device and medium

Abstract

The application discloses a kind of energy storage converter low voltage ride through control parameter identification method, system and medium, the method includes: to carry out multiple operating conditions low voltage ride through test to energy storage converter, and the measured data including transient interval, steady interval are collected;Parameter trajectory sensitivity is calculated to energy storage converter low voltage ride through full interval, according to sensitivity result, to be identified parameter is classified, and multiple objective optimization function is constructed in combination with the interval characteristics of measured data;Improved bayesian optimization algorithm is constructed;Using step-by-step identification strategy, improved bayesian optimization algorithm is called to carry out iterative optimization;Deviation index is calculated using multi-objective optimization function, whether the current parameter identification accuracy meets the preset engineering requirement is judged, and finally the identification result of energy storage converter low voltage ride through control parameter is output.The application can realize accurate and efficient identification of low voltage ride through control parameter under multiple operating conditions, improve the accuracy and engineering applicability of electromechanical transient simulation model of energy storage system.

Description

Technical Field

[0001] This invention relates to a method, device, and medium for identifying low-voltage ride-through control parameters of an energy storage converter, belonging to the field of power electronic converter parameter identification technology. Background Technology

[0002] With the continuous expansion of grid connection scale of new energy sources such as wind power and photovoltaics, battery energy storage systems have been widely used due to their advantages in smoothing out power output fluctuations from new energy sources and supporting grid voltage stability. As the core equipment for interaction between energy storage systems and the grid, the low voltage ride-through (LVRT) capability of the energy storage converter directly determines the grid-connected operation stability of the energy storage system under fault conditions. Therefore, establishing an accurate electromechanical transient simulation model of the energy storage converter is an important prerequisite for carrying out grid simulation, planning, and safety verification.

[0003] Currently, most energy storage equipment manufacturers keep the low-voltage ride-through (LVRT) control parameters and internal topology of energy storage converters confidential. The parameters in general simulation models deviate significantly from the actual equipment parameters, easily leading to distorted LVRT simulation results and consequently misjudgments in power grid planning. At present, the industry mainly uses least squares, particle swarm optimization, and differential evolution algorithms to identify converter control parameters. However, traditional numerical algorithms struggle to improve identification accuracy in multi-parameter, strongly coupled identification scenarios due to the nonlinear characteristics of the system. Conventional intelligent optimization algorithms suffer from poor adaptability to operating conditions, low iteration efficiency, and insufficient multi-objective optimization capabilities. Furthermore, most existing research does not differentiate parameter sensitivity, uniformly identifying all parameters, further increasing computational complexity. Simultaneously, it is difficult to balance identification accuracy across the transient and steady-state LVRT intervals, failing to meet the parameter identification needs under complex operating conditions such as multiple fault types, multiple voltage drop depths, and varying active power outputs. Summary of the Invention

[0004] To address the problems of insufficient accuracy in identifying low-voltage ride-through control parameters, poor adaptability to complex operating conditions, and low efficiency in iterative optimization in the existing technologies for energy storage converters, this invention provides a method, device, computer equipment, and storage medium for identifying low-voltage ride-through control parameters for energy storage converters. By combining semi-physical measured data, parameter sensitivity analysis, and a step-by-step identification strategy, it can achieve accurate and efficient identification of low-voltage ride-through control parameters under multiple operating conditions, thereby improving the accuracy and engineering applicability of the electromechanical transient simulation model of the energy storage system.

[0005] The first objective of this invention is to provide a method for identifying low-voltage ride-through control parameters of an energy storage converter.

[0006] The second objective of this invention is to provide a device for identifying low voltage ride-through control parameters of an energy storage converter.

[0007] A third objective of this invention is to provide a computer device.

[0008] A fourth objective of this invention is to provide a computer-readable storage medium.

[0009] The first objective of this invention can be achieved by adopting the following technical solution:

[0010] A method for identifying low-voltage ride-through control parameters of an energy storage converter, the method comprising:

[0011] Multi-condition low-voltage ride-through tests were conducted on the energy storage converter, and measured data including transient and steady-state ranges were collected to construct an identification benchmark dataset.

[0012] The parameter trajectory sensitivity of the low voltage ride-through of the energy storage converter is calculated across the entire range. Based on the sensitivity results, the parameters to be identified are divided into high-sensitivity parameters and low-sensitivity parameters. A multi-objective optimization function is constructed by combining the interval characteristics of the measured data.

[0013] An improved Bayesian optimization algorithm is constructed, which uses a Bayesian neural network as a surrogate model and a constrained multi-objective expectation improvement as the acquisition function.

[0014] A step-by-step identification strategy is adopted, first identifying high-sensitivity parameters and then identifying low-sensitivity parameters, and an improved Bayesian optimization algorithm is called to carry out iterative optimization.

[0015] Based on the preliminary identification results of iterative optimization, the deviation index is calculated using a multi-objective optimization function to determine whether the current parameter identification accuracy meets the preset engineering requirements. If yes, the identification results of the low voltage ride-through control parameters of the energy storage converter are output; otherwise, iterative optimization is continued until the parameter identification accuracy meets the preset engineering requirements, and the identification results of the low voltage ride-through control parameters of the energy storage converter are output.

[0016] Furthermore, the formula for calculating trajectory sensitivity is as follows:

[0017] ;

[0018] in, The number of sampling points; For the first One parameter to be identified; This refers to the output value of the BPA power system simulation software. For the first The amount of change in each input parameter; for and The ratio of .

[0019] Furthermore, the construction of a multi-objective optimization function based on the interval characteristics of the measured data includes:

[0020] If the parameter to be identified is a high-sensitivity parameter, the average deviation between the simulation data and the measured data of the active current and reactive current in the low voltage crossover steady-state region is selected as the optimization target.

[0021] If the parameter to be identified is a low-sensitivity parameter, the average deviation between the simulated and measured data of active and reactive currents in each segment of the low-voltage transient interval is selected as the optimization target.

[0022] Furthermore, the formula for calculating the average deviation is as follows:

[0023] ;

[0024] in, Per-unit values ​​for simulation data; To identify parameters; These are the per-unit values ​​of the measured data; and Calculate the first and last sequence numbers of the simulation data within the interval containing the deviation; and Calculate the first and last serial numbers of the measured data within the interval containing the deviation.

[0025] Furthermore, the construction of the improved Bayesian optimization algorithm, which uses a Bayesian neural network as the surrogate model and constrained multi-objective expectation improvement as the acquisition function, includes:

[0026] A Bayesian neural network with an attention mechanism is used as a surrogate model, and variational inference is used to approximate the posterior distribution of the network weights. The approximate posterior distribution formula is as follows:

[0027] ;

[0028] in, For variational parameters; The mean; Standard deviation; For unconstrained parameters, satisfy ; These are sampled values ​​of the weight parameters; This represents the total number of weights in the Bayesian neural network.

[0029] KL divergence is introduced for optimization, and attention branching is combined to achieve adaptive weight allocation for multiple objectives.

[0030] Based on the prediction results of the surrogate model, a single-objective expectation improvement function is constructed as follows:

[0031] ;

[0032] in, Let be the expected improvement value of the i-th objective function; The optimal value of the objective function in the training set; , These represent the mean and standard deviation predicted by the surrogate model, respectively. The cumulative distribution function of the standard normal distribution; It is the probability density function of the standard normal distribution;

[0033] By fusing attention weights to obtain the improved overall expectation across multiple objectives, and combining this with the constraint feasibility probability and constraint penalty term, the final constrained acquisition function is constructed, as shown below:

[0034] ;

[0035] in, Attention-weighted multi-objective expectation improves the overall sum; Let be the probability of feasibility for all constraints; This is the penalty coefficient; To constrain the degree of violation.

[0036] Furthermore, the step of calculating the deviation index using a multi-objective optimization function based on the preliminary identification results of iterative optimization, and determining whether the current parameter identification accuracy meets the preset engineering requirements, includes:

[0037] The preliminary identification results of iterative optimization are converted into simulation data. The simulation data is compared with the measured data. The deviation index is calculated using the average deviation formula of the multi-objective optimization function. At the same time, the constraint violation of Euclidean norm is introduced to comprehensively evaluate the current parameter identification accuracy.

[0038] If the current average deviation or the constraint violation Euclidean norm is greater than the corresponding preset threshold, it is determined that the current parameter identification accuracy does not meet the engineering requirements.

[0039] If the current average deviation and constraint violation Euclidean norm are less than or equal to the corresponding preset threshold, then the current parameter identification accuracy is determined to meet the engineering requirements.

[0040] Furthermore, the formula for calculating the violation of the Euclidean norm by the constraint is as follows:

[0041] ;

[0042] in, The average deviation of the current in each interval; Preset a deviation threshold for the corresponding interval; To constrain violations of the Euclidean norm.

[0043] The second objective of this invention can be achieved by adopting the following technical solution:

[0044] A device for identifying low-voltage ride-through control parameters of an energy storage converter, the device comprising:

[0045] The data acquisition module is used to conduct low-voltage ride-through tests on energy storage converters under multiple operating conditions, collect measured data including transient and steady-state ranges, and build an identification benchmark dataset.

[0046] The first construction module is used to calculate the parameter trajectory sensitivity of the low voltage ride-through of the energy storage converter across the entire range. Based on the sensitivity results, the parameters to be identified are divided into high-sensitivity parameters and low-sensitivity parameters. A multi-objective optimization function is constructed by combining the interval characteristics of the measured data.

[0047] The second building module is used to build an improved Bayesian optimization algorithm that uses a Bayesian neural network as a surrogate model and a constrained multi-objective expectation improvement as the acquisition function.

[0048] The iterative optimization module is used to employ a step-by-step identification strategy of first identifying high-sensitivity parameters and then identifying low-sensitivity parameters, and to call the improved Bayesian optimization algorithm to carry out iterative optimization.

[0049] The identification result output module is used to calculate the deviation index using a multi-objective optimization function based on the preliminary identification results of iterative optimization, and to determine whether the current parameter identification accuracy meets the preset engineering requirements. If yes, it outputs the identification results of the low voltage ride-through control parameters of the energy storage converter; if not, it continues to perform iterative optimization until the parameter identification accuracy meets the preset engineering requirements, and then outputs the identification results of the low voltage ride-through control parameters of the energy storage converter.

[0050] The third objective of this invention can be achieved by adopting the following technical solution:

[0051] A computer device includes a processor and a memory for storing a processor-executable program, wherein when the processor executes the program stored in the memory, it implements the above-described method for identifying low-voltage ride-through control parameters of an energy storage converter.

[0052] The fourth objective of this invention can be achieved by adopting the following technical solution:

[0053] A computer-readable storage medium storing a program that, when executed by a processor, implements the above-described method for identifying low-voltage ride-through control parameters of an energy storage converter.

[0054] The present invention has the following advantages over the prior art:

[0055] 1. This invention is based on a hardware-in-the-loop simulation platform to collect real multi-condition measured data, which fits the actual operating characteristics of the equipment on site. It ensures the basic accuracy of parameter identification from the data source level and can be adapted to complex engineering conditions such as three-phase short circuit, two-phase short circuit, and different voltage drop depths.

[0056] 2. This invention introduces parameter trajectory sensitivity calculation, classifies parameters into high and low sensitivity categories and adopts a step-by-step identification strategy, avoiding computational redundancy caused by simultaneous identification of all parameters, effectively reducing the computational load of the algorithm and shortening the iteration time.

[0057] 3. This invention uses a Bayesian neural network as a surrogate model, combined with an improved Bayesian optimization algorithm that improves the acquisition function with constrained multi-objective expectation. Compared with traditional particle swarm optimization, differential evolution and other algorithms, it has stronger multi-objective optimization ability, faster convergence speed and higher stability of identification results.

[0058] 4. This invention sets optimization targets for different regions of low voltage ride-through transient and steady state, taking into account the identification effect of the entire operating range, which greatly improves the overall identification accuracy of the LVRT control parameters of the energy storage converter, effectively optimizes the modeling effect of the electromechanical transient simulation model of the energy storage system, and has good engineering promotion value. Attached Figure Description

[0059] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0060] Figure 1 This is a flowchart of the low-voltage ride-through control parameter identification method for energy storage converters according to Embodiment 1 of the present invention.

[0061] Figure 2 The diagram below shows the grid-connected structure and control principle of the energy storage converter, which is an embodiment of the present invention.

[0062] Figure 3 This is a schematic diagram illustrating the division of the low-voltage ride-through transient and steady-state regions in Embodiment 1 of the present invention.

[0063] Figure 4 This is a structural block diagram of the low voltage ride-through control parameter identification device for energy storage converter according to Embodiment 2 of the present invention.

[0064] Figure 5 This is a structural block diagram of the computer device according to Embodiment 3 of the present invention. Detailed Implementation

[0065] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0066] Example 1:

[0067] like Figure 1 As shown in the figure, this embodiment provides a method for identifying low-voltage ride-through control parameters of an energy storage converter. The method includes the following steps:

[0068] S101. Conduct low-voltage ride-through tests on the energy storage converter under multiple operating conditions, collect measured data including transient and steady-state intervals, and construct an identification benchmark dataset.

[0069] This embodiment utilizes the RT-LAB hardware-in-the-loop simulation platform and the BPA power system electromechanical transient simulation software to jointly build a testing and identification environment. The object under test is a grid-connected energy storage converter. This embodiment focuses on data acquisition throughout the entire low-voltage ride-through (LVRT) process of the energy storage converter, dividing the LVRT operation into transient and steady-state intervals, and collecting time-series data of electrical quantities such as active current and reactive current throughout the process.

[0070] This embodiment sets up multiple typical test conditions, covering common faults and operating states in the power grid: the fault types include three-phase short circuit faults and two-phase short circuit faults; the voltage drop depth is set to 0.20pu, 0.50pu, and 0.75pu; the active power output of the energy storage converter is set to 0.2pu and 1.0pu.

[0071] All operating condition tests were completed using the RT-LAB hardware-in-the-loop simulation platform. Measured data were collected for the low-voltage ride-through transient range and steady-state range, and a benchmark dataset for identification was constructed.

[0072] S102. Perform parameter trajectory sensitivity calculations for the low-voltage ride-through of the energy storage converter across the entire range. Based on the sensitivity results, divide the parameters to be identified into high-sensitivity parameters and low-sensitivity parameters, and construct a multi-objective optimization function by combining the range characteristics of the measured data.

[0073] This embodiment uses BPA power system simulation software to build a single-unit electromechanical transient model of the energy storage converter. The control structure of the energy storage converter in the BPA software is as follows: Figure 2 As shown, it mainly includes a battery pack, inverter and control system, LCL filter circuit, and transformer, etc.; among them, the control system collects the three-phase AC voltage at the grid connection point. Three-phase alternating current The active power is obtained through coordinate transformation and electrical quantity calculation. reactive power and the positive sequence voltage component at the grid connection point The system determines the converter's operating status (power output mode, fault condition, etc.). The control system generates a reactive current reference value through power / voltage outer loop regulation. Active current reference value It is then transmitted to the inner current loop, outputting converter control commands.

[0074] This embodiment employs trajectory sensitivity analysis to perform full-range sensitivity calculations on all LVRT control parameters of the energy storage converter in the BPA model, quantifying the influence of each parameter on transient and steady-state electrical outputs. The trajectory sensitivity calculation formula is as follows:

[0075] ;

[0076] in, The number of sampling points; For the first One parameter to be identified; This refers to the output value of the BPA power system simulation software. For the first The amount of change in each input parameter; for and The ratio of .

[0077] The specific calculation process for parameter sensitivity is as follows: First, set non-zero typical values ​​for all parameters to be identified; set a three-phase short-circuit fault at the grid connection point of the energy storage converter, causing the positive sequence voltage at the grid connection point to drop to 0.5 pu, the fault duration to be set to 1.214 s, and the sampling time to be set to 0.002 s, recording the active and reactive currents throughout the fault process as system output data; during the calculation process, keep the other identification parameters unchanged, only increasing the value of a single parameter to be identified by 5%, and calculate the trajectory sensitivity of each parameter one by one according to the formula. The low-voltage ride-through process of the energy storage converter is as follows: Figure 3 As shown, the low-voltage ride-through process can be divided into transient regions. , and steady-state range , , The calculated results of parameter trajectory sensitivity for the steady-state and transient regions of low-voltage fault crossing are shown in Table 1.

[0078] Table 1. Calculation results of trajectory sensitivity in each section during the low-voltage ride-through process.

[0079]

[0080] Based on the magnitude of the trajectory sensitivity, all parameters to be identified are divided into two categories: high-sensitivity parameters and low-sensitivity parameters. High-sensitivity parameters include... , , , , , , , Low sensitivity parameters include , , , Based on the equipment manual and engineering experience, reasonable value ranges for all parameters to be identified were defined.

[0081] If the parameter to be identified is a high-sensitivity parameter, the average deviation between the simulated and measured data of active and reactive currents during the low-voltage ride-through steady-state region is selected as the optimization target; if the parameter to be identified is a low-sensitivity parameter, the average deviation between the simulated and measured data of active and reactive currents in each segment of the low-voltage ride-through transient region is selected as the optimization target; the formula for calculating the average deviation is as follows:

[0082] ;

[0083] in, Per-unit values ​​for simulation data; To identify parameters; These are the per-unit values ​​of the measured data; and Calculate the first and last sequence numbers of the simulation data within the interval containing the deviation; and Calculate the first and last serial numbers of the measured data within the interval containing the deviation.

[0084] S103. Construct an improved Bayesian optimization algorithm with a Bayesian neural network as the surrogate model and a constrained multi-objective expectation improvement as the acquisition function.

[0085] This embodiment addresses the multi-objective, constrained black-box optimization problem of LVRT parameter identification in energy storage converters by constructing an improved Bayesian optimization algorithm. This algorithm uses a Bayesian neural network (BNN) as a surrogate model for the Bayesian optimization framework, and designs a constrained multi-objective expectation improvement function as the algorithm's acquisition function.

[0086] This embodiment uses a Bayesian neural network with an attention mechanism as a surrogate model, and uses variational inference to approximate the posterior distribution of the network weights. The approximate posterior distribution formula is as follows:

[0087] ;

[0088] in, For variational parameters; The mean; Standard deviation; For unconstrained parameters, satisfy ; These are sampled values ​​of the weight parameters; This represents the total number of weights in the Bayesian neural network.

[0089] To measure the difference between the approximate posterior and the true posterior, KL divergence is introduced for optimization, and attention branching is combined to achieve adaptive weight allocation for multiple objectives.

[0090] Based on the prediction results of the surrogate model, a single-objective expectation improvement function is constructed as follows:

[0091] ;

[0092] in, Let be the expected improvement value of the i-th objective function; The optimal value of the objective function in the training set; , These represent the mean and standard deviation predicted by the surrogate model, respectively. The cumulative distribution function of the standard normal distribution; It is the probability density function of the standard normal distribution;

[0093] By fusing attention weights to obtain the improved overall expectation across multiple objectives, and combining this with the constraint feasibility probability and constraint penalty term, the final constrained acquisition function is constructed, as shown below:

[0094] ;

[0095] in, Attention-weighted multi-objective expectation improves the overall sum; Let be the probability of feasibility for all constraints; This is the penalty coefficient; To constrain the degree of violation.

[0096] During the algorithm iteration process, by maximizing the acquisition function The optimal candidate parameter samples are selected in each round to complete the adaptive selection of the optimization samples.

[0097] S104. A step-by-step identification strategy is adopted, first identifying high-sensitivity parameters and then identifying low-sensitivity parameters, and an improved Bayesian optimization algorithm is called to carry out iterative optimization.

[0098] Based on the parameter sensitivity division results in step S102, this embodiment adopts a step-by-step identification strategy of first identifying high-sensitivity parameters and then identifying low-sensitivity parameters, and calls the improved Bayesian optimization algorithm constructed in step S103 to carry out iterative optimization.

[0099] The first stage is high-sensitivity parameter identification: defining S102... , , , All low-sensitivity parameters are set to typical engineering values ​​and fixed, only for , , , , , , , Eight high-sensitivity parameters were optimized, with a maximum of 320 iterations in this stage. The parameter combinations output from each iteration were imported into the BPA power system simulation model, and the BPA energy storage converter electromechanical transient simulation model was run to simulate low-voltage ride-through under full operating conditions, outputting simulation data such as active current and reactive current.

[0100] After iterating through the high-sensitivity parameters, the optimal parameter combination is obtained. This set of parameters is then fixed throughout the process, and the second stage of identification begins. The second stage is the low-sensitivity parameter identification: while keeping the high-sensitivity parameters unchanged, the low-sensitivity parameters are further refined. , , , Four low-sensitivity parameters were iteratively optimized, with a maximum of 40 iterations allowed in this stage. After two rounds of step-by-step iterations were completed, preliminary identification results of the low-voltage ride-through control parameters of the energy storage converter were obtained.

[0101] S105. Based on the preliminary identification results of iterative optimization, the deviation index is calculated using a multi-objective optimization function to determine whether the current parameter identification accuracy meets the preset engineering requirements.

[0102] This embodiment imports the preliminary identification results from iterative optimization into the BPA power system simulation software to reproduce all test conditions, including three-phase short circuit, two-phase short circuit, different voltage drop depths, and different active power outputs, and outputs simulation data. The simulation data is then compared with the measured data collected by the RT-LAB hardware-in-the-loop platform. The average deviation formula in S102 is used to calculate the error index, and constraints violating the Euclidean norm are introduced to comprehensively evaluate the current parameter identification accuracy. The Euclidean norm calculation formula is as follows:

[0103] ;

[0104] in, This represents the average deviation of the current in each interval; Preset a deviation threshold for the corresponding interval; To constrain violations of the Euclidean norm.

[0105] In this embodiment, the judgment thresholds are set in accordance with national standards and engineering application standards: the average deviation thresholds for active and reactive currents during low-voltage ride-through steady-state intervals are both set to 0.05; the average deviation thresholds for each current during low-voltage ride-through transient intervals are both set to 0.05.

[0106] Calculations show that the preliminary identification results correspond to the following indicators: average deviation of active current in the low-voltage ride-through steady-state range is 2.4273%, and average deviation of reactive current is 1.9202%; the average deviation of each current in the low-voltage ride-through transient range is distributed between 4.3243% and 7.2254%. Based on the above indicators, the constraint violation of the Euclidean norm is calculated. All calculation results are compared with preset thresholds one by one to determine whether the current parameter identification accuracy meets the engineering requirements.

[0107] If the current average deviation or the constraint violation Euclidean norm is greater than the corresponding preset threshold, it is determined that the current parameter identification accuracy does not meet the engineering requirements, and the process returns to step S104 to continue a new round of iterative optimization.

[0108] If the current average deviation and constraint violation Euclidean norm are less than or equal to the corresponding preset threshold, it is determined that the current parameter identification accuracy meets the engineering requirements, the entire iteration process is terminated, and the identification results of the low voltage ride-through control parameters of the energy storage converter are output.

[0109] After a finite number of rounds of supplementary iterations, all error indicators in this embodiment are reduced to within the preset threshold, the iteration is terminated, and the final identified parameters can be directly used for the calibration of the electromechanical transient simulation model of the energy storage converter.

[0110] It should be noted that although the method operations of the above embodiments are described in a specific order, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. On the contrary, the described steps may be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0111] Example 2:

[0112] like Figure 4 As shown, this embodiment provides a low-voltage ride-through control parameter identification device for an energy storage converter. The device includes a data acquisition module 401, a first construction module 402, a second construction module 403, an iterative optimization module 404, and an identification result output module 405. The specific descriptions of each module are as follows:

[0113] The data acquisition module 401 is used to conduct multi-condition low-voltage ride-through tests on the energy storage converter, collect measured data including transient and steady-state intervals, and construct an identification benchmark dataset.

[0114] The first construction module 402 is used to perform parameter trajectory sensitivity calculation on the low voltage ride-through range of the energy storage converter. Based on the sensitivity results, the parameters to be identified are divided into high-sensitivity parameters and low-sensitivity parameters, and a multi-objective optimization function is constructed in combination with the range characteristics of the measured data.

[0115] The second building module 403 is used to build an improved Bayesian optimization algorithm that uses a Bayesian neural network as a surrogate model and a constrained multi-objective expectation improvement as the acquisition function.

[0116] The iterative optimization module 404 is used to adopt a step-by-step identification strategy of first identifying high-sensitivity parameters and then identifying low-sensitivity parameters, and to call the improved Bayesian optimization algorithm to carry out iterative optimization.

[0117] The identification result output module 405 is used to calculate the deviation index using a multi-objective optimization function based on the preliminary identification results of iterative optimization, and to determine whether the current parameter identification accuracy meets the preset engineering requirements. If yes, it outputs the identification results of the low voltage ride-through control parameters of the energy storage converter; if no, it continues to perform iterative optimization until the parameter identification accuracy meets the preset engineering requirements, and then outputs the identification results of the low voltage ride-through control parameters of the energy storage converter.

[0118] The specific implementation of each module in this embodiment can be found in Embodiment 1 above, and will not be repeated here. It should be noted that the device provided in this embodiment is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure can be divided into different functional modules to complete all or part of the functions described above.

[0119] Example 3:

[0120] like Figure 5 As shown, this embodiment provides a computer device, which includes a processor 502, a memory, an input device 503, a display device 504, and a network interface 505 connected via a system bus 501. The processor 502 provides computing and control capabilities. The memory includes a non-volatile storage medium 506 and internal memory 507. The non-volatile storage medium 506 stores an operating system, computer programs, and a database. The internal memory 507 provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium 506. When the computer program is executed by the processor 502, it implements the low-voltage ride-through control parameter identification method for energy storage converters described in Embodiment 1, as follows:

[0121] Multi-condition low-voltage ride-through tests were conducted on the energy storage converter, collecting measured data including transient and steady-state intervals to construct an identification benchmark dataset. Parameter trajectory sensitivity calculations were performed across the entire low-voltage ride-through range of the energy storage converter. Based on the sensitivity results, the parameters to be identified were divided into high-sensitivity and low-sensitivity parameters, and a multi-objective optimization function was constructed based on the interval characteristics of the measured data. An improved Bayesian optimization algorithm was constructed, using a Bayesian neural network as the surrogate model and constrained multi-objective expectation improvement as the acquisition function. A step-by-step identification strategy was adopted, first identifying high-sensitivity parameters and then low-sensitivity parameters, and the improved Bayesian optimization algorithm was invoked for iterative optimization. Based on the preliminary identification results of the iterative optimization, the deviation index was calculated using the multi-objective optimization function to determine whether the current parameter identification accuracy meets the preset engineering requirements. If yes, the identification results of the low-voltage ride-through control parameters of the energy storage converter were output; otherwise, iterative optimization was continued until the parameter identification accuracy meets the preset engineering requirements, and the identification results of the low-voltage ride-through control parameters of the energy storage converter were output.

[0122] Example 4:

[0123] This embodiment provides a computer-readable storage medium storing a computer program. When the program is executed by a processor, the processor executes the computer program stored in the memory to implement the low-voltage ride-through control parameter identification method for energy storage converters described in Embodiment 1 above, as follows:

[0124] Multi-condition low-voltage ride-through tests were conducted on the energy storage converter, collecting measured data including transient and steady-state intervals to construct an identification benchmark dataset. Parameter trajectory sensitivity calculations were performed across the entire low-voltage ride-through range of the energy storage converter. Based on the sensitivity results, the parameters to be identified were divided into high-sensitivity and low-sensitivity parameters, and a multi-objective optimization function was constructed based on the interval characteristics of the measured data. An improved Bayesian optimization algorithm was constructed, using a Bayesian neural network as the surrogate model and constrained multi-objective expectation improvement as the acquisition function. A step-by-step identification strategy was adopted, first identifying high-sensitivity parameters and then low-sensitivity parameters, and the improved Bayesian optimization algorithm was invoked for iterative optimization. Based on the preliminary identification results of the iterative optimization, the deviation index was calculated using the multi-objective optimization function to determine whether the current parameter identification accuracy meets the preset engineering requirements. If yes, the identification results of the low-voltage ride-through control parameters of the energy storage converter were output; otherwise, iterative optimization was continued until the parameter identification accuracy meets the preset engineering requirements, and the identification results of the low-voltage ride-through control parameters of the energy storage converter were output.

[0125] It should be noted that the computer-readable storage medium in this embodiment can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor device, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0126] In this embodiment, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used or combined with an instruction execution device, apparatus, or device. In this embodiment, the computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable program. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable storage medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use or combined with an instruction execution device, apparatus, or device. The computer program contained on the computer-readable storage medium can be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination thereof.

[0127] The computer-readable storage medium described above can be used to write computer programs for executing this embodiment in one or more programming languages ​​or combinations thereof. These programming languages ​​include object-oriented programming languages—such as Java, Python, and C++—and conventional procedural programming languages—such as C or similar programming languages. The program can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0128] In summary, this invention combines semi-physical measured data, parameter sensitivity analysis, and a step-by-step identification strategy to achieve accurate and efficient identification of low-voltage ride-through control parameters under multiple operating conditions. This improves the accuracy and engineering applicability of the electromechanical transient simulation model of the energy storage system and solves the problems of insufficient identification accuracy of low-voltage ride-through control parameters, poor adaptability to complex operating conditions, and low efficiency of iterative optimization in the prior art for energy storage converters.

[0129] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope disclosed in the present invention, based on the technical solution and inventive concept of the present invention, shall fall within the scope of protection of the present invention.

Claims

1. A method for identifying low-voltage ride-through control parameters of an energy storage converter, characterized in that, The method includes: Multi-condition low-voltage ride-through tests were conducted on the energy storage converter, and measured data including transient and steady-state ranges were collected to construct an identification benchmark dataset. The parameter trajectory sensitivity of the low voltage ride-through of the energy storage converter is calculated across the entire range. Based on the sensitivity results, the parameters to be identified are divided into high-sensitivity parameters and low-sensitivity parameters. A multi-objective optimization function is constructed by combining the interval characteristics of the measured data. An improved Bayesian optimization algorithm is constructed, which uses a Bayesian neural network as a surrogate model and a constrained multi-objective expectation improvement as the acquisition function. A step-by-step identification strategy is adopted, first identifying high-sensitivity parameters and then identifying low-sensitivity parameters, and an improved Bayesian optimization algorithm is called to carry out iterative optimization. Based on the preliminary identification results of iterative optimization, the deviation index is calculated using a multi-objective optimization function to determine whether the current parameter identification accuracy meets the preset engineering requirements. If yes, the identification results of the low voltage ride-through control parameters of the energy storage converter are output; otherwise, iterative optimization is continued until the parameter identification accuracy meets the preset engineering requirements, and the identification results of the low voltage ride-through control parameters of the energy storage converter are output.

2. The method for identifying low-voltage ride-through control parameters of an energy storage converter according to claim 1, characterized in that, The formula for calculating trajectory sensitivity is as follows: ； in, The number of sampling points; For the first One parameter to be identified; This is the output value of the BPA power system simulation software. For the first The amount of change in each input parameter; for and The ratio of .

3. The method for identifying low-voltage ride-through control parameters of an energy storage converter according to claim 1, characterized in that, The multi-objective optimization function constructed by combining the interval features of measured data includes: If the parameter to be identified is a high-sensitivity parameter, the average deviation between the simulation data and the measured data of the active current and reactive current in the low voltage crossover steady-state region is selected as the optimization target. If the parameter to be identified is a low-sensitivity parameter, the average deviation between the simulated and measured data of active and reactive currents in each segment of the low-voltage transient interval is selected as the optimization target.

4. The method for identifying low-voltage ride-through control parameters of an energy storage converter according to claim 3, characterized in that, The formula for calculating the average deviation is as follows: ； in, Per-unit values ​​for simulation data; To identify parameters; These are the per-unit values ​​of the measured data; and Calculate the first and last sequence numbers of the simulation data within the interval containing the deviation; and Calculate the first and last serial numbers of the measured data within the interval containing the deviation.

5. The method for identifying low-voltage ride-through control parameters of an energy storage converter according to claim 1, characterized in that, The improved Bayesian optimization algorithm, which uses a Bayesian neural network as a surrogate model and constrained multi-objective expectation improvement as the acquisition function, includes: A Bayesian neural network with an attention mechanism is used as a surrogate model, and variational inference is used to approximate the posterior distribution of the network weights. The approximate posterior distribution formula is as follows: ； in, For variational parameters; The mean; Standard deviation; For unconstrained parameters, satisfy ; These are sampled values ​​of the weight parameters; This represents the total number of weights in the Bayesian neural network. KL divergence is introduced for optimization, and attention branching is combined to achieve adaptive weight allocation for multiple objectives. Based on the prediction results of the surrogate model, a single-objective expectation improvement function is constructed as follows: ； in, Let be the expected improvement value of the i-th objective function; The optimal value of the objective function in the training set; , These represent the mean and standard deviation predicted by the surrogate model, respectively. The cumulative distribution function of the standard normal distribution; It is the probability density function of the standard normal distribution; By fusing attention weights to obtain the improved overall expectation across multiple objectives, and combining this with the constraint feasibility probability and constraint penalty term, the final constrained acquisition function is constructed, as shown below: ； in, Attention-weighted multi-objective expectation improves the overall sum; Let be the probability of feasibility for all constraints; This is the penalty coefficient; To constrain the degree of violation.

6. The method for identifying low-voltage ride-through control parameters of an energy storage converter according to claim 1, characterized in that, The step of calculating the deviation index using a multi-objective optimization function based on the preliminary identification results of iterative optimization, and determining whether the current parameter identification accuracy meets the preset engineering requirements, includes: The preliminary identification results of iterative optimization are converted into simulation data. The simulation data is compared with the measured data. The deviation index is calculated using the average deviation formula of the multi-objective optimization function. At the same time, the constraint violation of Euclidean norm is introduced to comprehensively evaluate the current parameter identification accuracy. If the current average deviation or the constraint violation Euclidean norm is greater than the corresponding preset threshold, it is determined that the current parameter identification accuracy does not meet the engineering requirements. If the current average deviation and constraint violation Euclidean norm are less than or equal to the corresponding preset threshold, then the current parameter identification accuracy is determined to meet the engineering requirements.

7. The method for identifying low-voltage ride-through control parameters of an energy storage converter according to claim 6, characterized in that, The formula for calculating the violation of the Euclidean norm by the constraint is as follows: ； in, The average deviation of the current in each interval; Preset a deviation threshold for the corresponding interval; To constrain violations of the Euclidean norm.

8. A device for identifying low-voltage ride-through control parameters of an energy storage converter, characterized in that, The device includes: The data acquisition module is used to conduct low-voltage ride-through tests on energy storage converters under multiple operating conditions, collect measured data including transient and steady-state ranges, and build an identification benchmark dataset. The first construction module is used to calculate the parameter trajectory sensitivity of the low voltage ride-through of the energy storage converter across the entire range. Based on the sensitivity results, the parameters to be identified are divided into high-sensitivity parameters and low-sensitivity parameters. A multi-objective optimization function is constructed by combining the interval characteristics of the measured data. The second building module is used to build an improved Bayesian optimization algorithm that uses a Bayesian neural network as a surrogate model and a constrained multi-objective expectation improvement as the acquisition function. The iterative optimization module is used to employ a step-by-step identification strategy of first identifying high-sensitivity parameters and then identifying low-sensitivity parameters, and to call the improved Bayesian optimization algorithm to carry out iterative optimization. The identification result output module is used to calculate the deviation index using a multi-objective optimization function based on the preliminary identification results of iterative optimization, and to determine whether the current parameter identification accuracy meets the preset engineering requirements. If yes, it outputs the identification results of the low voltage ride-through control parameters of the energy storage converter; if not, it continues to perform iterative optimization until the parameter identification accuracy meets the preset engineering requirements, and then outputs the identification results of the low voltage ride-through control parameters of the energy storage converter.

9. A computer device comprising a processor and a memory for storing a processor-executable program, characterized in that, When the processor executes the program stored in the memory, it implements the low voltage ride-through control parameter identification method for energy storage converters as described in any one of claims 1-7.

10. A computer-readable storage medium storing a program, characterized in that, When the program is executed by the processor, it implements the method for identifying low voltage ride-through control parameters of energy storage converters as described in any one of claims 1-7.

Images