A filter equivalent circuit design simulation system suitable for energy storage converter

Abstract

The application relates to the field of power electronics, in particular to a filter equivalent circuit design simulation system suitable for an energy storage converter, which comprises a main circuit topology unit, a control strategy unit, a parameter optimization and design unit, a simulation engine unit and a performance evaluation unit, wherein: the main circuit topology unit is used for building a filter equivalent circuit topology of the energy storage converter and configuring passive element parameters of inductance, capacitance and resistance; the filter parameters are automatically and highly robustly optimized through genetic algorithm-particle swarm hybrid optimization, multi-objective collaborative optimization and digital twin auxiliary optimization; dynamic working condition evaluation, reliability modeling and artificial intelligence auxiliary analysis are combined to form a "design-optimization-verification-iteration" closed loop; the harmonic suppression capacity, the efficiency and the long-term operation reliability of the filter are significantly improved; the design cycle is shortened and the development cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of power electronics, and in particular to a filter equivalent circuit design simulation system suitable for an energy storage converter. Background Art

[0002] Power storage converter (PCS) filters are key passive components installed between the converter and the grid or load. They are primarily used to filter the harmonic currents and voltages generated during converter operation, improving power quality and ensuring safe and stable converter operation and compliance with grid access standards. However, conventional power storage converter filter design relies on manual experience and presents the following prominent issues:

[0003] Low parameter matching efficiency: Manual debugging makes it difficult to take into account multiple constraints such as harmonic attenuation rate, efficiency, and volume. As a result, parameter matching can only achieve local optimization, resulting in insufficient harmonic suppression capability and slow dynamic response.

[0004] Lack of high-frequency modeling: The lack of accurate modeling for parasitic parameters such as inductor skin effect and capacitor dielectric loss resulted in a 20% error between simulation results and actual hardware testing, requiring repeated hardware iterations. This resulted in a design cycle of 2-4 weeks and increased costs by over 30%.

[0005] Insufficient reliability assessment: Performance was evaluated only through static operating condition testing, without the introduction of reliability modeling methods such as Markov chains. This made it impossible to predict the impact of factors such as component aging and temperature fluctuations on filter life, resulting in a high failure rate in actual operation.

[0006] Insufficient intelligence: Relying on manual analysis of design defects, lacking intelligent optimization capabilities based on historical data, unable to quickly generate parameter adjustment suggestions, and low design iteration efficiency.

[0007] Based on this, the present invention provides a filter equivalent circuit design simulation system suitable for energy storage converters to solve the above-mentioned technical problems. Summary of the Invention

[0008] The purpose of the present invention is to provide a filter equivalent circuit design simulation system suitable for energy storage converters. The present invention uses genetic algorithm-particle swarm hybrid optimization, multi-objective collaborative optimization and digital twin assisted optimization to achieve automated and highly robust optimization of filter parameters. Combined with dynamic operating condition evaluation, reliability modeling and artificial intelligence-assisted analysis, a "design-optimization-verification-iteration" closed loop is formed, which significantly improves the harmonic suppression capability, efficiency and long-term operation reliability of the filter, shortens the design cycle and reduces development costs.

[0009] To achieve the above object, the present invention provides the following technical solutions:

[0010] The present invention provides a filter equivalent circuit design simulation system suitable for energy storage converters, including a main circuit topology unit, a control strategy unit, a parameter optimization and design unit, a simulation engine unit, and a performance evaluation unit, wherein:

[0011] The main circuit topology unit is used to build the filter equivalent circuit topology of the energy storage converter and configure the passive component parameters of the inductor, capacitor, and resistor;

[0012] The control strategy unit is used to realize the coordinated control of the filter and the converter, including proportional resonance control and active damping algorithm;

[0013] The parameter optimization and design unit: Based on the multi-objective constraints of harmonic attenuation rate and efficiency, it automatically calculates the optimal parameters of the filter with high robustness through genetic algorithm-particle swarm hybrid optimization, multi-objective collaborative optimization and digital twin assisted optimization;

[0014] The simulation engine unit is used to perform time domain and frequency domain simulations to verify the performance of the filter under dynamic conditions;

[0015] The performance evaluation unit is used to quantify key indicators such as total harmonic distortion, efficiency, and dynamic response through dynamic operating condition evaluation, reliability modeling, and artificial intelligence-assisted analysis, and to evaluate its long-term operating reliability and comprehensively judge the effectiveness of the design.

[0016] The main circuit topology unit includes a topology building module, a component parameter configuration module, and a parasitic parameter modeling module, wherein:

[0017] The topology building module is used to generate the filter topology framework of LCL and LLCL;

[0018] The component parameter configuration module is used to set the nominal parameters and parasitic parameters of passive components such as inductance, capacitance and resistance;

[0019] The parasitic parameter modeling module is used to consider the high-frequency characteristics of the components and establish an equivalent circuit model including parasitic resistance and capacitance.

[0020] The control strategy unit includes a proportional resonance control module, an active damping algorithm module, and a collaborative control adaptation module, wherein:

[0021] The proportional resonance control module is used to achieve multi-band adjustment and set the resonance frequency and gain parameters for specific subharmonics;

[0022] The active damping algorithm module is used to include two modes: capacitor current feedback virtual resistance damping and grid voltage feedforward phase compensation damping;

[0023] The collaborative control adaptation module is used to automatically match the optimal control strategy according to the filter topology type to achieve collaborative operation with the converter.

[0024] The parameter optimization and design unit includes a hybrid optimization algorithm module, a multi-objective collaboration module, and a digital twin auxiliary module, wherein:

[0025] The hybrid optimization algorithm module is used to generate an initial solution space by using a genetic algorithm global search combined with a particle swarm optimization local fine-tuning;

[0026] The multi-objective collaborative module: based on the multi-objective constraints of harmonic attenuation rate and efficiency, balances and optimizes various parameters;

[0027] The digital twin auxiliary module is used to construct a digital twin model, correct simulation parameters in combination with real-time data, and calibrate optimization results.

[0028] The digital twin auxiliary module constructs a digital twin model, combines real-time data to correct simulation parameters, and calibrates optimization results. The specific operations are as follows:

[0029] A1: Physical system data acquisition: Real-time acquisition of voltage, current, temperature and loss data during the operation of the energy storage converter filter;

[0030] A2: Virtual simulation model construction: Based on the main circuit topology unit, a high-fidelity simulation model consistent with the physical system is built;

[0031] A3: Online parameter correction: Dynamically adjust the inductor and capacitor component parameters in the simulation model based on the actual operating data collected;

[0032] A4: Optimization feedback: Feedback the modified model to the hybrid optimization algorithm module to iteratively calibrate the filter parameter optimization results.

[0033] The online parameter correction in A3 updates the filter parameters according to the error minimization objective function:

[0034]

[0035] in, , Output voltage and current for the simulation model. , These are actual system measurements.

[0036] The simulation engine unit includes a time domain simulation module, a frequency domain simulation module, and a fault injection module, wherein:

[0037] The time domain simulation module is used to simulate dynamic operating condition sequences of load mutations and power grid faults to verify the time domain performance of the filter;

[0038] The frequency domain simulation module is used to analyze the amplitude-frequency and phase-frequency characteristics of the filter and evaluate the harmonic suppression capability;

[0039] The fault injection module is used to accelerate simulation calculations using a parallel architecture.

[0040] The performance evaluation unit includes a dynamic working condition evaluation module, a reliability modeling module, and an artificial intelligence analysis module, wherein:

[0041] The dynamic working condition evaluation module is used to quantify the total harmonic distortion rate, efficiency, and dynamic response indicators in the dynamic process;

[0042] The reliability modeling module is used to predict the failure probability of the filter throughout its life cycle using a Markov chain model;

[0043] The artificial intelligence analysis module trains AI models based on historical data, automatically identifies design defects and provides optimization suggestions.

[0044] The artificial intelligence analysis module trains an AI model based on historical data, automatically identifies design flaws and provides optimization suggestions. The specific operations are as follows:

[0045] B1: Data preprocessing:

[0046] ① Extract filter design parameters, simulation results and corresponding design defect labels from the historical database;

[0047] ② Perform Z-score normalization on continuous parameters and one-hot encoding on discrete parameters;

[0048] ③ Use the SMOTE algorithm to oversample the unbalanced samples to balance the proportion of various design defect samples;

[0049] B2: Feature Engineering:

[0050] ① Calculate the correlation matrix between parameters and screen out features that are highly correlated with design defects;

[0051] ② Constructing time series features to identify potential defects under transient conditions;

[0052] B3: Model training steps:

[0053] ① Use 5-fold cross validation to divide the preprocessed data into training set and test set;

[0054] ② Initialize the deep neural network model, set the input layer dimension to the number of features, use a 3-layer ReLU activation function for the hidden layer, and use a Softmax classifier for the output layer;

[0055] ③ Use the Adam optimizer to minimize the cross entropy loss function, with a batch size of 64, a training round of 100, and an early stopping strategy of patience = 10;

[0056] B4: Defect Identification:

[0057] ① Input the filter design data to be evaluated into the trained model to obtain the predicted probability of various design defects;

[0058] ② If the predicted probability of a certain type of defect exceeds the threshold, the defect is determined to exist and the specific circuit component is located;

[0059] B5: Generate optimization suggestions:

[0060] ① Based on the defect type, retrieve historical success cases from the knowledge base;

[0061] ② If there is no matching case in the knowledge base, a reinforcement learning algorithm is used to generate new suggestions, with the objective function of maximizing system performance improvement;

[0062] ③ Perform simulation verification on the generated suggestions to ensure that the optimized design meets the performance indicators.

[0063] The knowledge base in B5 is constructed in the following way:

[0064] C1: Extract the design parameters, working condition constraints, and optimization measures from historical successful cases;

[0065] C2: Use knowledge graph technology to establish parameter-defect-solution association rules;

[0066] C3: Trigger recommendations based on case similarity matching.

[0067] Compared with the prior art, the present invention has the following beneficial effects:

[0068] The present invention realizes automated and highly robust optimization of filter parameters through genetic algorithm-particle swarm hybrid optimization, multi-objective collaborative optimization and digital twin-assisted optimization, and combines dynamic operating condition evaluation, reliability modeling and artificial intelligence-assisted analysis to form a "design-optimization-verification-iteration" closed loop, significantly improving the filter's harmonic suppression capability, efficiency and long-term operating reliability, shortening the design cycle and reducing development costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 This is a system diagram of a filter equivalent circuit design simulation system suitable for energy storage converters according to the present invention.

[0070] Figure 2 This is a flow chart of digital twin parameter calibration in a filter equivalent circuit design simulation system suitable for an energy storage converter according to the present invention.

[0071] Figure 3 This is a flowchart of AI defect diagnosis in a filter equivalent circuit design simulation system suitable for energy storage converters according to the present invention.

[0072] Figure 4 This is a control strategy selection flow chart for a filter equivalent circuit design simulation system suitable for an energy storage converter according to the present invention.

[0073] Description of Figure Numbers:

[0074] 100. Main circuit topology unit; 101. Topology building module; 102. Component parameter configuration module; 103. Parasitic parameter modeling module; 200. Control strategy unit; 201. Proportional resonance control module; 202. Active damping algorithm module; 203. Collaborative control adaptation module; 300. Parameter optimization and design unit; 301. Hybrid optimization algorithm module; 302. Multi-objective collaborative module; 303. Digital twin auxiliary module; 400. Simulation engine unit; 401. Time domain simulation module; 402. Frequency domain simulation module; 403. Fault injection module; 500. Performance evaluation unit; 501. Dynamic working condition evaluation module; 502. Reliability modeling module; 503. Artificial intelligence analysis module. DETAILED DESCRIPTION

[0075] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0076] Example:

[0077] like Figures 1-4As shown, this embodiment provides a filter equivalent circuit design simulation system for energy storage converters, including a main circuit topology unit 100, a control strategy unit 200, a parameter optimization and design unit 300, a simulation engine unit 400, and a performance evaluation unit 500, wherein: the main circuit topology unit 100 is used to build the filter equivalent circuit topology of the energy storage converter and configure the passive component parameters of the inductor, capacitor, and resistor; the control strategy unit 200 is used to realize the coordinated control of the filter and the converter, including proportional resonance control and active damping algorithm; the parameter optimization and Design unit 300: Based on the multi-objective constraints of harmonic attenuation rate and efficiency, the optimal parameters of the filter with high robustness are automatically calculated through genetic algorithm-particle swarm hybrid optimization, multi-objective collaborative optimization and digital twin assisted optimization; Simulation engine unit 400: Used to perform time domain and frequency domain simulations to verify the performance of the filter under dynamic working conditions; Performance evaluation unit 500: Used to quantify the key indicators of total harmonic distortion rate, efficiency, and dynamic response through dynamic working condition evaluation, reliability modeling and artificial intelligence-assisted analysis, and evaluate its long-term operating reliability to comprehensively judge the effectiveness of the design.

[0078] It should be noted that the filter model constructed by the main circuit topology unit 100 is optimized and controlled by the control strategy unit 200, and the optimal parameters are iteratively generated by the parameter optimization and design unit 300, the simulation engine unit 400 verifies the dynamic performance, and finally the defect and reliability data are fed back to the preceding unit through the performance evaluation unit 500.

[0079] In this embodiment, it should also be noted that the main circuit topology unit 100 includes a topology building module 101, a component parameter configuration module 102, and a parasitic parameter modeling module 103, wherein: the topology building module 101 is used to generate the filter topology structure framework of LCL and LLCL; the component parameter configuration module 102 is used to set the nominal parameters and parasitic parameters of the passive components of inductance, capacitance, and resistance; the parasitic parameter modeling module 103 is used to consider the high-frequency characteristics of the components and establish an equivalent circuit model including parasitic resistance and capacitance.

[0080] It should be noted that the LCL and LLCL topology frameworks generated by the topology building module 101 are set with nominal parameters by the component parameter configuration module 102, and then high-frequency parasitic effects are injected by the parasitic parameter modeling module 103, and gradually improved into high-precision equivalent circuit models.

[0081] Furthermore, it should be noted that a graphical modeling tool for the standard filter structure of LCL and LLCL is provided, and users can quickly build filter topologies by dragging and dropping. The component parameter configuration module 102 supports two parameter setting modes: manual input and database retrieval. Users can manually input the nominal parameters of inductance, capacitance, and resistance, and set the parasitic parameters of ESR, ESL, and tanδ at the same time. The parasitic parameter modeling module 103 is based on high-frequency circuit theory and constructs an equivalent circuit model for high-frequency characteristics such as the skin effect and capacitor dielectric loss of the inductor coil. The specific operations are as follows: Construction of the skin effect equivalent circuit model of the inductor coil: When the inductor coil works at high frequency, the current tends to flow on the surface of the conductor, that is, the skin effect. This phenomenon will cause the equivalent resistance of the inductor to increase, and the inductance value will also change. In order to construct an inductor equivalent circuit model that takes into account the skin effect, we can regard the inductor coil as consisting of multiple thin layers. At high frequencies, the current is mainly concentrated in the thin layer on the surface of the conductor. The following methods can be used for modeling:

[0082] 1. Lumped parameter model: The inductor coil is equivalent to a series combination of a resistor and an inductor, where the resistor takes into account the increase in equivalent resistance caused by the skin effect. It can be calculated by the following formula: ,in, is the DC resistance, is the skin depth, a is the conductor radius;

[0083] 2. Distributed parameter model: For more accurate modeling, the inductor can be considered as consisting of multiple small segments of resistance and inductance, each of which takes into account the skin effect. This model can be implemented using transmission line theory or finite element analysis.

[0084] Constructing a Capacitor Dielectric Loss Equivalent Circuit Model: When a capacitor operates at high frequencies, dielectric loss cannot be ignored. Dielectric loss is energy loss caused by polarization in the capacitor dielectric. To construct a capacitor equivalent circuit model that accounts for dielectric loss, the capacitor is typically equated to a parallel combination of an ideal capacitor and a resistor, where the resistor represents dielectric loss. The equivalent resistance, R, can be calculated using the following formula: ,in, is the angular frequency, C is the ideal capacitance value, is the dielectric loss tangent.

[0085] In this embodiment, it should also be noted that the control strategy unit 200 includes a proportional resonance control module 201, an active damping algorithm module 202, and a collaborative control adaptation module 203, wherein: the proportional resonance control module 201 is used to realize multi-band adjustment and set the resonant frequency and gain parameters for specific subharmonics; the active damping algorithm module 202 is used to include two modes: capacitor current feedback virtual resistance damping and grid voltage feedforward phase compensation damping; the collaborative control adaptation module 203 is used to automatically match the optimal control strategy according to the filter topology type to achieve collaborative operation with the converter.

[0086] It should be noted that the proportional resonance control module 201 and the active damping algorithm module 202 provide harmonic suppression and system stabilization functions respectively. The two control strategies are dynamically integrated through the collaborative control adaptation module 203, and the optimal control mode is intelligently matched according to the filter topology. After the two control strategies of proportional resonance control and active damping algorithm are dynamically integrated through the collaborative control adaptation module 203, a Bode diagram is generated to analyze the phase margin of the system. When the phase margin is >45°, it is confirmed that the current control strategy can meet the system stability requirements; if the phase margin is ≤45°, the gain of the proportional resonance controller is adjusted, and the Bode diagram is regenerated for analysis until the phase margin is >45°.

[0087] Furthermore, it should be noted that the proportional resonance control module 201 adopts a multi-band PR controller to set the corresponding resonant frequency and gain for specific harmonics (such as the 3rd, 5th, and 7th) to achieve precise harmonic suppression. The active damping algorithm module 202 provides two damping control modes: ① Capacitor current feedback virtual resistance damping: suppressing the resonance spike by injecting virtual damping current; ② Grid voltage feedforward phase compensation damping: improving the response characteristics of the filter to grid disturbances. The collaborative control adaptation module 203 automatically matches the optimal control strategy based on the filter topology type through a preset rule base. The rule base refers to a logical set that maps the filter topology type to the most appropriate control strategy based on expert knowledge, engineering experience, and a large amount of simulation or experimental data; 1. Implementation process:

[0088] Step 1: Input Feature Extraction

[0089] Static parameters: filter type (LCL / LLCL), inductance / capacitance value, parasitic parameters;

[0090] Dynamic parameters: real-time grid harmonic spectrum, load fluctuation rate, temperature;

[0091] Step 2: Rule base matching

[0092] Primary screening: Extract candidate strategy sets from the rule base based on topology type and working condition labels; Performance prediction: Call the digital twin model to simulate candidate strategies and output predicted values ​​such as THD and dynamic response time;

[0093] Step 3: Strategy Scoring and Selection

[0094] Scoring formula:

[0095]

[0096] Output: Select the strategy and parameter combination with the highest total score;

[0097] Step 4: Closed-loop verification and iteration

[0098] Real-time verification: injecting real data into the simulation engine to verify the effectiveness of the strategy;

[0099] Rule base update: If verification fails, record the new case and optimize the rule weights;

[0100] 2. Rule base content:

[0101]

[0102] 3. Output results:

[0103] Control strategy: proportional resonance control parameters, active damping mode and parameters;

[0104] Performance indicators: expected THD, efficiency, stability margin;

[0105] Exception handling: If the match fails, the manual intervention flag is triggered and an alternative solution is output;

[0106] 4. Notes:

[0107] Rule base maintenance: Rules need to be regularly supplemented based on new cases (such as new grid disturbance scenarios);

[0108] Real-time requirements: The matching cycle must be less than 10ms to ensure the response speed of dynamic working conditions.

[0109] In this embodiment, it should also be noted that the parameter optimization and design unit 300 includes a hybrid optimization algorithm module 301, a multi-objective collaboration module 302, and a digital twin auxiliary module 303, wherein: the hybrid optimization algorithm module 301: is used to generate an initial solution space by using a genetic algorithm global search combined with a particle swarm optimization local fine-tuning; the multi-objective collaboration module 302: is used to balance and optimize various parameters based on the multi-objective constraints of harmonic attenuation rate and efficiency; the digital twin auxiliary module 303: is used to correct simulation parameters and calibrate optimization results by constructing a digital twin model and combining real-time data. The specific operations are as follows: A1: Physical system data acquisition: by acquiring the voltage, current, temperature and loss data of the energy storage converter filter in real time; A2: Virtual simulation model construction: based on the main circuit topology unit 100, a high-fidelity simulation model consistent with the physical system is built; A3: Parameter online correction: dynamically adjust the component parameters of the inductor and capacitor in the simulation model according to the actual operation data collected; the parameter online correction in A3 updates the filter parameters according to the error minimization objective function:

[0110]

[0111] in, , Output voltage and current for the simulation model. , A4: Optimization feedback: Feedback the modified model to the hybrid optimization algorithm module 301 to iteratively calibrate the filter parameter optimization results.

[0112] It should be noted that the hybrid optimization algorithm module 301 generates the initial parameter solution space. After multi-constraint optimization is performed by the multi-objective collaboration module 302, the digital twin auxiliary module 303 realizes dynamic parameter calibration through a closed-loop process of "data acquisition-model construction-parameter correction-feedback iteration".

[0113] Furthermore, it should be noted that the genetic algorithm in the hybrid optimization algorithm module 301 searches for potential solution space globally through selection, crossover, and mutation operations; the particle swarm optimization uses the optimal solution of the genetic algorithm as the initial population, and quickly converges to the global optimal solution through information sharing between particles and local search. The multi-objective collaboration module 302 constructs a multi-objective optimization model based on the Pareto frontier, incorporates objective functions such as harmonic attenuation rate, efficiency, volume, and cost into a unified framework, and uses the weighted summation method or the ε-constraint method to transform the multi-objective problem into a single-objective optimization problem. By adjusting the weight coefficient to balance the priorities of different objectives, the optimal parameter solution set that meets the engineering requirements is output. Specific implementation methods: 1. Optimization problem modeling:

[0114] Unified expression of objective function:

[0115]

[0116] Constraints:

[0117] (Avoid grid frequency interference);

[0118] Inductor saturation current ;

[0119] Capacitor withstand voltage ;

[0120] 2. Weight coefficient setting rules

[0121] Project priority and weight correspondence table:

[0122] 3. Specific Examples

[0124] Case Background:

[0125] Design an LCL filter for photovoltaic grid connection, the requirements are:

[0126] ①THD (IEEE 519 standard);

[0127] ② Efficiency ;

[0128] ③ Volume ;

[0129] ④ Cost ;

[0130] Implementation steps:

[0131] Step 1: Parameter initialization

[0132] Policy variables: (The value range is limited by the component database);

[0133] Initial weights: (prioritize THD and efficiency);

[0134] Step 2: Multi-objective optimization process

[0135] 1. Genetic Algorithm (Global Search):

[0136] Population size = 100, iteration 50 generations, generate Pareto frontier;

[0137] Typical solution examples:

[0138] Solution A: THD = 2.8%, Efficiency = 97.2%, Volume = 0.048 m³, Cost = $210;

[0139] Solution B: THD = 2.5%, Efficiency = 96.8%, Volume = 0.055 m³, Cost = $190;

[0140] 2. Dynamic weight adjustment:

[0141] If the user requires strict cost , then the weight is adjusted to , re-optimize to get solution C:

[0142] THD = 2.9%, efficiency = 97.0%, volume = 0.052m³, cost = $198;

[0143] Step 3: Digital Twin Verification

[0144] Input the solution C parameters into the simulation model, and the dynamic working condition test results are:

[0145] THD instantaneous peak value during load step = 3.1% (slightly exceeding the standard);

[0146] Feedback to optimization module for fine-tuning The target will be reached after a 10% increase;

[0147] 4. Output parameter solution example

[0148]

[0149] In A1, the voltage, current, temperature and loss data of the energy storage converter filter during operation are obtained through sensors, with a sampling frequency of not less than 10kHz.

[0150] In this embodiment, it should also be noted that the simulation engine unit 400 includes a time domain simulation module 401, a frequency domain simulation module 402, and a fault injection module 403, wherein: the time domain simulation module 401 is used to simulate a dynamic operating condition sequence of load mutations and power grid faults to verify the time domain performance of the filter; the frequency domain simulation module 402 is used to analyze the amplitude-frequency and phase-frequency characteristics of the filter and evaluate the harmonic suppression capability; the fault injection module 403 is used to accelerate simulation calculations using a parallel architecture.

[0151] It should be noted that the time domain simulation module 401 and the frequency domain simulation module 402 verify the filter performance from the dimensions of dynamic operating response and frequency characteristic analysis respectively, and the fault injection module 403 provides computational acceleration support for the former two through a parallel architecture.

[0152] Further, it needs to be explained that the time domain simulation module 401 supports custom working condition sequence editing, and the user can set the trigger time, duration and amplitude change of dynamic events such as load step, grid voltage drop and frequency deviation. The frequency domain simulation module 402 generates the amplitude-frequency characteristic curve and phase-frequency characteristic curve of the filter through sweep analysis, and automatically labels key indicators such as cutoff frequency, resonance frequency and gain margin. Support multiple visualization forms such as Bode diagram and Nyquist diagram, which is convenient for users to analyze the stability and harmonic suppression ability of the filter. The fault injection module 403 adopts GPU parallel computing architecture, and decomposes the simulation task into multiple sub-tasks for parallel processing.

[0153] In this embodiment, it should also be noted that the performance evaluation unit 500 includes a dynamic working condition evaluation module 501, a reliability modeling module 502, and an artificial intelligence analysis module 503, wherein: the dynamic working condition evaluation module 501: is used to quantify the total harmonic distortion rate, efficiency, and dynamic response indicators in the dynamic process; the reliability modeling module 502: is used to use the Markov chain model to predict the failure probability of the filter throughout its life cycle; the artificial intelligence analysis module 503: is used to train the AI ​​model based on historical data, automatically identify design defects and provide optimization suggestions. The specific operations are as follows: B1: Data preprocessing: ① Extract filter design parameters, simulation results and corresponding design defect labels from the historical database; ② Perform Z-score normalization on continuous parameters and perform one-hot encoding on discrete parameters; ③ Use the SMOTE algorithm to oversample the unbalanced samples to balance the proportion of samples of various design defects; B2: Feature Engineering: ① Calculate the correlation matrix between parameters to filter out features that are highly correlated with design defects; ② Construct time series features to identify potential defects under transient conditions; B3: Model Training Steps: ① Use 5-fold cross-validation to divide the preprocessed data into training and test sets; ② Initialize the deep neural network model, set the input layer dimension to the number of features, use a 3-layer ReLU activation function in the hidden layer, and use a Softmax classifier in the output layer; ③ Use the Adam optimizer Minimize the cross-entropy loss function, set the batch size to 64, the number of training rounds to 100, and the early stopping strategy patience = 10; B4: Defect Identification: ① Input the filter design data to be evaluated into the trained model to obtain the predicted probability of each type of design defect; ② If the predicted probability of a certain defect type exceeds a threshold, the defect is determined to be present and the specific circuit component is located; B5: Optimization Recommendation Generation: ① Based on the defect type, retrieve historical successful cases from the knowledge base; ② If there are no matching cases in the knowledge base, generate new recommendations using a reinforcement learning algorithm, with the objective function of maximizing system performance improvement; ③ Simulate and verify the generated recommendations to ensure that the optimized design meets the performance indicators. The knowledge base in B5 is constructed through the following methods: C1: Extracting design parameter, operating condition constraint, and optimization measure triplets from historical successful cases; C2: Using knowledge graph technology to establish parameter-defect-solution association rules; C3: Triggering recommendations based on case similarity matching.

[0154] It should be noted that, in the performance evaluation unit 500, the dynamic working condition evaluation module 501 and the reliability modeling module 502 output evaluation data from the dimensions of real-time performance indicator quantification and full life cycle reliability prediction respectively, and the artificial intelligence analysis module 503 is based on the knowledge base constructed by the data of the first two and historical cases, and realizes intelligent identification of design defects and recommendation of optimization strategies through full-process AI analysis including data preprocessing, feature engineering, model training, defect identification and generation of optimization suggestions.

[0155] Furthermore, it should be noted that B1 includes design parameters such as inductance and capacitance; simulation results such as THD and efficiency; B2 includes screening out features highly correlated with design defects such as L / C ratio and resonant frequency; B4 includes exceeding a threshold such as 0.7; specific circuit components such as "the grid-side inductance parameters of the LCL filter are unreasonable"; and B5 includes retrieving historical successful cases from the knowledge base such as "when THD exceeds the standard, it is recommended to increase the grid-side inductance by 15%."

[0156] like Figures 1-4 As shown, this embodiment provides a filter equivalent circuit design and simulation system suitable for energy storage converters. The specific usage method is as follows: first, open the system interface, select the filter type (LCL / LLCL) in the main circuit topology unit 100, build the topology by dragging component icons or directly call the preset template, enter the nominal parameters (inductance value, capacitance value, etc.) in the component parameter configuration module 102, or call typical parameters from the database, enable the parasitic parameter modeling module 103, set the high-frequency characteristic parameters (such as ESR and skin depth coefficient), and then enter the control strategy unit 200 to select the control mode according to the topology type: LCL filter: activate the proportional resonance control module 201 (set the 3rd / 5th / 7th harmonic resonance frequency) and the active damping algorithm module 202 (select the capacitor current feedback mode), LLCL filter: add grid voltage feedforward phase compensation, verify the strategy compatibility through the collaborative control adaptation module 203, the system automatically generates a Bode diagram to verify the stability, and enters the parameter optimization stage after it is correct. Set the optimization target (such as THD <3%, efficiency > 98%) in the parameter optimization and design unit 300, start the hybrid optimization algorithm module 301 to generate the initial parameter solution, the multi-objective collaborative module 302 outputs the Pareto optimal solution set, and then connect to the actual converter hardware. Perform closed-loop calibration through the digital twin auxiliary module 303: first collect voltage / current data in real time; then compare the simulation output with the measured data, automatically correct the inductor / capacitor parameters; finally iterate until the error After convergence (e.g., RMSE < 1%), the test conditions are further configured in the simulation engine unit 400: time domain simulation module 401: set load step (0%-100%) and grid voltage drop (20%); frequency domain simulation module 402: scan the 50Hz-10kHz frequency band to generate the impedance characteristic curve, enable the fault injection module 403 to accelerate the simulation, and obtain the dynamic response waveform. Finally, the dynamic working condition evaluation module 501 automatically calculates indicators such as THD and efficiency; the reliability modeling module 502 outputs the MTBF prediction value (e.g., >100,000 hours), and the artificial intelligence analysis module 503 performs defect diagnosis: input simulation data, and the model annotates potential defects (e.g., "resonance peak shift"); the system recommends optimization solutions (e.g., "increase the damping resistance by 15%"), adjusts the design according to the suggestions, and returns to the parameter optimization stage for re-optimization until all indicators meet the standards.

[0157] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0158] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the present invention to the specific embodiments described. Obviously, many modifications and variations are possible based on the content of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A filter equivalent circuit design simulation system suitable for energy storage converter, characterized in that: It includes a main circuit topology unit (100), a control strategy unit (200), a parameter optimization and design unit (300), a simulation engine unit (400), and a performance evaluation unit (500), wherein: The main circuit topology unit (100) is used to construct a filter equivalent circuit topology of the energy storage converter and configure the passive component parameters of the inductor, capacitor, and resistor; The control strategy unit (200) is used to realize the coordinated control of the filter and the converter, including proportional resonance control and active damping algorithm; The parameter optimization and design unit (300) automatically calculates the optimal parameters of the filter with high robustness based on the multi-objective constraints of harmonic attenuation rate and efficiency through genetic algorithm-particle swarm hybrid optimization, multi-objective collaborative optimization and digital twin assisted optimization; The simulation engine unit (400) is used to perform time domain and frequency domain simulations to verify the performance of the filter under dynamic working conditions; The performance evaluation unit (500) is used to quantify the key indicators of total harmonic distortion rate, efficiency, and dynamic response through dynamic working condition evaluation, reliability modeling, and artificial intelligence-assisted analysis, and to evaluate its long-term operating reliability and comprehensively judge the effectiveness of the design; The main circuit topology unit (100) comprises a topology building module (101), a component parameter configuration module (102), and a parasitic parameter modeling module (103), wherein: The topology building module (101) is used to generate the filter topology framework of LCL and LLCL; The component parameter configuration module (102) is used to set the nominal parameters and parasitic parameters of passive components such as inductance, capacitance and resistance; The parasitic parameter modeling module (103) is used to consider the high-frequency characteristics of the component and establish an equivalent circuit model including parasitic resistance and capacitance; The control strategy unit (200) includes a proportional resonance control module (201), an active damping algorithm module (202), and a collaborative control adaptation module (203), wherein: The proportional resonance control module (201) is used to implement multi-band regulation and set the resonance frequency and gain parameters for specific subharmonics; The active damping algorithm module (202) is configured to include two modes: capacitor current feedback virtual resistance damping and grid voltage feedforward phase compensation damping; The collaborative control adaptation module (203) is used to automatically match the optimal control strategy according to the filter topology type to achieve collaborative operation with the converter; In the digital twin auxiliary module (303), a digital twin model is constructed, simulation parameters are corrected in combination with real-time data, and the optimization results are calibrated. The specific operations are as follows: A1: Physical system data acquisition: Real-time acquisition of voltage, current, temperature and loss data during the operation of the energy storage converter filter; A2: Construction of virtual simulation model: Building a high-fidelity simulation model consistent with the physical system based on the main circuit topology unit (100); A3: Online parameter correction: Dynamically adjust the inductor and capacitor component parameters in the simulation model based on the actual operating data collected; A4: Optimization feedback: feeding back the modified model to the hybrid multi-channel range-Doppler spectrum optimization algorithm module (301) to iteratively calibrate the filter parameter optimization results; The simulation engine unit (400) includes a time domain simulation module (401), a frequency domain simulation module (402), and a fault injection module (403), wherein: The time domain simulation module (401) is used to simulate a dynamic operating condition sequence of load mutation and power grid fault, and verify the time domain performance of the filter; The frequency domain simulation module (402) is used to analyze the amplitude-frequency and phase-frequency characteristics of the filter and evaluate the harmonic suppression capability; The fault injection module (403) is used to accelerate simulation calculations using a parallel architecture.

2. A filter equivalent circuit design simulation system suitable for an energy storage converter according to claim 1, characterized in that: The parameter optimization and design unit (300) includes a hybrid optimization algorithm module (301), a multi-objective collaboration module (302), and a digital twin auxiliary module (303), wherein: The hybrid optimization algorithm module (301) is used to generate an initial solution space by using a genetic algorithm global search combined with a particle swarm optimization local fine tuning; The multi-objective collaborative module (302) balances and optimizes various parameters based on the multi-objective constraints of harmonic attenuation rate and efficiency; The digital twin auxiliary module (303) is used to construct a digital twin model, correct simulation parameters in combination with real-time data, and calibrate optimization results.

3. The filter equivalent circuit design simulation system for an energy storage converter according to claim 1, characterized in that: The online parameter correction in A3 updates the filter parameters according to the error minimization objective function: ， in, , Output voltage and current for the simulation model. , These are actual system measurements.

4. The filter equivalent circuit design simulation system for an energy storage converter according to claim 1, characterized in that: The performance evaluation unit (500) comprises a dynamic working condition evaluation module (501), a reliability modeling module (502), and an artificial intelligence analysis module (503), wherein: The dynamic working condition evaluation module (501) is used to quantify the total harmonic distortion rate, efficiency, and dynamic response indicators in the dynamic process; The reliability modeling module (502) is used to predict the failure probability of the filter throughout its life cycle using a Markov chain model; The artificial intelligence analysis module (503) is used to train an AI model based on historical data, automatically identify design defects and provide optimization suggestions.

5. A filter equivalent circuit design simulation system suitable for an energy storage converter according to claim 4, characterized in that: The artificial intelligence analysis module (503) trains an AI model based on historical data, automatically identifies design defects and gives optimization suggestions. The specific operations are as follows: B1: Data preprocessing: ① Extract filter design parameters, simulation results and corresponding design defect labels from the historical database; ② Perform Z-score normalization on continuous parameters and one-hot encoding on discrete parameters; ③ Use the SMOTE algorithm to oversample the unbalanced samples to balance the proportion of various design defect samples; B2: Feature Engineering: ① Calculate the correlation matrix between parameters and screen out features that are highly correlated with design defects; ② Constructing time series features to identify potential defects under transient conditions; B3: Model training steps: ① Use 5-fold cross validation to divide the preprocessed data into training set and test set; ② Initialize the deep neural network model, set the input layer dimension to the number of features, use a 3-layer ReLU activation function for the hidden layer, and use a Softmax classifier for the output layer; ③ Use the Adam optimizer to minimize the cross entropy loss function, with a batch size of 64, a training round of 100, and an early stopping strategy of patience = 10; B4: Defect Identification: ① Input the filter design data to be evaluated into the trained model to obtain the predicted probability of various design defects; ② If the predicted probability of a certain type of defect exceeds the threshold, the defect is determined to exist and the specific circuit component is located; B5: Generate optimization suggestions: ① Based on the defect type, retrieve historical success cases from the knowledge base; ② If there is no matching case in the knowledge base, a reinforcement learning algorithm is used to generate new suggestions, with the objective function of maximizing system performance improvement; ③ Perform simulation verification on the generated suggestions to ensure that the optimized design meets the performance indicators.

6. A filter equivalent circuit design simulation system suitable for an energy storage converter according to claim 5, characterized in that: The knowledge base in B5 is constructed in the following way: C1: Extract the design parameters, working condition constraints, and optimization measures from historical successful cases; C2: Use knowledge graph technology to establish parameter-defect-solution association rules; C3: Trigger recommendations based on case similarity matching.

Images