Active reconfiguration control system for dynamically balancing energy storage string

Abstract

The invention belongs to the technical field of energy storage string control, and discloses an active reconstruction control system for energy storage string dynamic equalization. The system comprises a battery unit health dynamic prediction module, a multi-target topological optimization algorithm module, a multi-target decision security verification module, a preventive management strategy planning module and an online calibration learning module, and calculates and predicts the health state of each battery unit in real time based on a preprocessed feature data set to obtain a battery health prediction report. The method comprises the steps of calculating an optimal string connection topology scheme based on a battery health prediction report, obtaining a multi-target optimization result report, obtaining a safety verification instruction report, formulating preventive maintenance suggestions based on the battery health prediction report, obtaining a long-term strategy suggestion report, and generating a model calibration report. The method has the remarkable advantages of being high in balance durability, high in multi-performance compatibility and good in battery model self-adaptive effect.

Description

Technical Field

[0001] This invention relates to the field of energy storage string control technology, and more specifically, to an active reconfiguration control system for dynamic balancing of energy storage strings. Background Technology

[0002] Active reconfiguration control system for dynamic string balancing is an advanced energy management technology applied to battery energy storage systems. Its core principle is to dynamically change the electrical connection topology between battery cells or modules (i.e., "string reconfiguration") through a controllable power electronic switching network, thereby optimizing the combination of battery cells with different performance parameters (such as capacity, internal resistance, and state of health) to achieve balanced distribution of energy and power at the system level, improve overall available capacity, extend system life, and enhance safety and economy.

[0003] However, traditional active reconfiguration control systems for dynamic balancing of energy storage strings suffer from the following shortcomings during use: First, traditional systems largely rely on data such as the current voltage or temperature of battery cells for reconfiguration decisions. This type of decision-making inherently has a lag, resulting in numerous reconfiguration operations and a short effective period. Furthermore, repeated switching operations increase system losses and the risk of failure. Second, traditional reconfiguration algorithms mostly focus on minimizing the current inter-string imbalance, which is a single-objective and single-time-scale optimization problem. This optimization approach easily leads to unnecessary and large-scale triggering of reconfiguration mechanisms in an attempt to achieve instantaneous balance. The topology switching of traditional systems sacrifices system reliability and economy for instantaneous absolute equilibrium, which in turn reduces the degree of future imbalance under long-term use. Third, the battery model parameters used in traditional systems are mostly fixed. However, the battery itself will undergo various negative changes with the increase of cycle number and operating environment, which will reduce the accuracy and effectiveness of reconfiguration decisions. In summary, how to effectively solve the problems of poor equilibrium persistence, limitations of instantaneous optimal reconfiguration strategy and insufficient battery model adaptability in traditional systems has become the bottleneck problem that active reconfiguration control system for dynamic equilibrium of energy storage strings needs to face and solve.

[0004] In view of this, the present invention proposes an active reconfiguration control system for dynamic balancing of energy storage strings to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution, including: The data acquisition and feature extraction module is used to send data acquisition instructions to the BMS system based on the acquisition conditions, obtain raw measurement data, and perform preprocessing and feature extraction to obtain a preprocessed feature dataset. Furthermore, the steps of sending data acquisition commands to the BMS system based on the acquisition conditions to obtain raw measurement data, and performing preprocessing and feature extraction include: S1.1: Send data acquisition instructions to the BMS system based on the acquisition conditions to obtain the raw measurement dataset. The data acquisition instructions include high-frequency impedance spectrum measurement instructions and basic data acquisition instructions. S1.2: Based on the acquisition timestamp and according to the battery cells, all data items in the original measurement dataset are aligned to obtain the aligned dataset; S1.3: Based on the aligned dataset, the moving average filter is used for processing, and the 3σ method is used for outlier removal. At the same time, linear interpolation is used for padding to obtain the preprocessed dataset. S1.4: Based on the preprocessed dataset, feature extraction is performed to obtain a feature vector set. The features extracted include ohmic internal resistance, charge transfer resistance, diffusion impedance characteristic frequency, and phase angle characteristics. S1.5: Based on the feature vector set, a multi-dimensional feature vector is constructed for each battery cell to obtain a preprocessed feature dataset; S1.6: Store the preprocessed feature dataset in the database; The battery cell health dynamic prediction module is used to calculate and predict the health status of each battery cell in real time based on the preprocessed feature dataset, and obtain a battery health prediction report. Furthermore, the steps for real-time calculation and prediction of the health status of each battery cell based on the preprocessed feature dataset include: S2.1: Based on the database, retrieve the preprocessed feature dataset of all battery cells for the past L cycles to obtain the cycle feature sequence; S2.2: Based on the periodic feature sequence, the current health status of the battery is calculated using a pre-trained gradient boosting tree model to obtain a health assessment sequence. Simultaneously, the capacity decay rate is calculated to obtain a capacity decay sequence. The specific formula set for the calculation is as follows: ; The first one was obtained respectively Health assessment value of each battery cell and the Capacity decay rate of individual battery cells ,in, For the first Each battery cell at time point Preprocessed feature data, The set of all learnable parameters for the gradient boosting tree model. This is the cumulative number of loops; S2.3: Input the periodic feature sequence into the battery health prediction model and output the battery health prediction sequence; S2.4: Based on the battery health prediction sequence, the Monte Carlo Dropout method is used to evaluate the prediction uncertainty, and the prediction variance sequence and 95% confidence interval are obtained; S2.5: Package the health assessment sequence, capacity decay sequence, battery health prediction sequence, prediction variance sequence, and 95% confidence interval to obtain the battery health prediction report; S2.6: Output the battery health prediction report to the multi-objective topology optimization algorithm module and the preventive management strategy planning module respectively; The multi-objective topology optimization algorithm module is used to calculate the optimal string connection topology scheme based on the battery health prediction report and obtain a multi-objective optimization result report. Furthermore, the steps for calculating the optimal string connection topology based on the battery health prediction report include: S3.1: Based on the battery health prediction report, define the connection matrix and constraints, where the constraints are that each battery cell must be connected to one and only one string, and the number of batteries in each string is equal; S3.2: Define the objective function based on the battery health prediction report and the connection matrix to obtain a set of objective functions. The objective functions include minimizing the difference in battery health prediction values ​​within the group, minimizing the switching cost, and maximizing the consistency of battery health prediction values ​​within the group. S3.3: Use the NSGA-II algorithm to randomly generate U connection matrices that satisfy the constraints, and perform iterative optimization. When the target number of iterations is reached, obtain the Pareto front solution set. S3.4: For each Pareto solution, calculate the performance index and obtain the topology performance evaluation report. The performance index includes the objective function, the efficiency improvement evaluation value, and the number of switching operations. S3.5: Package the Pareto front solution set and topological performance evaluation report to obtain the multi-objective optimization result report; S3.6: Output the multi-objective optimization result report to the multi-objective decision security verification module; The multi-objective decision security verification module is used to select the final execution plan from the multi-objective optimization result report, perform security verification, and obtain a security verification instruction report; Furthermore, the steps for selecting the final implementation plan from the multi-objective optimization results report and performing security verification include: S4.1: Based on the long-term strategy guidance vector and combined with the current working conditions, the weights are adjusted to obtain the comprehensive weights; S4.2: Based on the Pareto front solution set in the multi-objective optimization result report, calculate the weighted score for each Pareto solution and select the solution with the highest score as the candidate solution; S4.3: Perform simulated safety verification based on candidate solutions. If the safety verification result is passed, proceed to step S4.4. If the safety verification result is failed, terminate the operation and generate a safety verification report to be output to the staff receiving end. The simulated safety verification includes electrical safety verification, thermal safety simulation and switching capability verification. S4.4: Compare the execution plan of the current connection with the candidate plan output in step S4.3 to generate a reconstruction instruction sequence; S4.5: Based on steps S4.1 to S4.4, generate a security verification report; S4.6: Package and reconstruct the instruction sequence and security verification report to obtain the security verification instruction report; S4.7: Output the security verification instruction report to the dynamic coordination instruction distribution module; The preventive management strategy planning module is used to formulate preventive maintenance recommendations based on battery health prediction reports, and obtain a long-term strategy recommendation report; Furthermore, the steps for developing preventative maintenance recommendations based on battery health prediction reports include: S5.1: Calculate the degradation rate of each battery cell based on the battery health prediction report to obtain the degradation rate sequence; S5.2: Based on the decay rate sequence, the clustering distance is calculated using the Ward method, and the battery cells are divided into C homogeneous groups to obtain the battery cluster set; S5.3: Based on battery clusters and according to the degradation threshold, battery clusters with degradation rates less than the degradation threshold are classified as fast degradation groups, and battery clusters with degradation rates greater than or equal to the degradation threshold are classified as stable groups. Corresponding strategies are formulated according to the differences between fast degradation groups and stable groups to obtain a long-term strategy report. S5.4: Generate long-term strategy guidance vectors based on the current system stage, which includes the initial, middle and final stages; S5.5: Based on steps S5.1 to S5.4, generate maintenance recommendations and obtain a preventive maintenance recommendation report; S5.6: Package the long-term strategy guidance vector and preventive maintenance recommendation report to obtain the long-term strategy recommendation report; S5.7: Output the long-term strategy recommendation report to the multi-objective decision security verification module; The dynamic coordination instruction distribution module is used to coordinate short-term and long-term strategies for security verification instruction reports and long-term strategy suggestion reports, generate the final execution instruction set, and perform instruction distribution and execution monitoring to obtain a coordination execution report. Furthermore, the steps of coordinating short-term and long-term strategies in the security verification instruction report and long-term strategy recommendation report to generate the final execution instruction set, and distributing and monitoring the execution of the instructions include: S6.1: Calculate the imbalance of the current system based on the security verification instruction report to obtain the reconstruction evaluation value. When the reconstruction evaluation value is less than the reconstruction threshold, skip the reconstruction. When the number of skips is greater than the maximum allowed number of skips, force the reconstruction to be triggered. S6.2: Based on the long-term strategy guidance vector and combined with the current working conditions, the optimal reconfiguration time is calculated to obtain the optimal reconfiguration time point; S6.3: Verify whether the reconstructed instruction sequence is feasible at the current time. If the verification result is not feasible, return to the multi-objective topology optimization algorithm module for recalculation. If the verification result is feasible, package the reconstructed instruction sequence and the execution time window to obtain the final execution instruction set, and proceed to step S6.4. S6.4: Distribute the final execution instruction set to the switch execution system, monitor the execution process, and obtain a coordinated execution report; S6.5: Output the coordinated execution report to the online calibration learning module; The online calibration learning module is used to compare the predictions and actual results, dynamically update the battery health prediction model, and generate a model calibration report. Furthermore, the steps of comparing the predicted and actual results and dynamically updating the battery health prediction model include: S7.1: Obtain the battery health prediction sequence based on the battery health prediction report, and calculate the prediction error value sequence by comparing it with the actual battery health value at the same time. S7.2: Calculate the overall prediction accuracy index based on the prediction error value sequence to obtain a prediction accuracy report. The prediction accuracy index includes the mean absolute error, root mean square error, and maximum error. S7.3: When the mean absolute error is greater than the mean absolute error threshold, generate a model calibration signal and proceed to step S7.4; S7.4: When a model calibration signal appears, construct a training dataset and adjust the battery health prediction model parameters to obtain a model calibration report; S7.5: Output the model calibration report to the battery cell health dynamic prediction module; Furthermore, an active reconfiguration control method for dynamic balancing of energy storage strings includes: S1: Based on the acquisition conditions, send a data acquisition command to the BMS system to obtain the raw measurement data, and perform preprocessing and feature extraction to obtain a preprocessed feature dataset; S2: Calculate and predict the health status of each battery cell in real time based on the preprocessed feature dataset to obtain a battery health prediction report; S3: Calculate the optimal string connection topology based on the battery health prediction report and obtain a multi-objective optimization result report; S4: Select the final execution plan from the multi-objective optimization result report, perform security verification, and obtain a security verification instruction report; S5: Based on the battery health prediction report, preventive maintenance recommendations are formulated to obtain a long-term strategy recommendation report; S6: Coordinate short-term and long-term strategies for security verification instruction reports and long-term strategy recommendation reports, generate the final execution instruction set, distribute and monitor the execution of instructions, and obtain a coordinated execution report; S7: Compare the predicted and actual results, dynamically update the battery health prediction model, and generate a model calibration report.

[0006] The technical effects and advantages of the active reconfiguration control system for dynamic balancing of energy storage strings in this invention are as follows: This invention sends data acquisition commands to the BMS system based on acquisition conditions to obtain raw measurement data. This data is then preprocessed and features extracted to obtain a preprocessed feature dataset. Based on this dataset, the health status of each battery cell is calculated and predicted in real time, resulting in a battery health prediction report. The optimal string connection topology is calculated based on the battery health prediction report, resulting in a multi-objective optimization result report. The final execution plan is selected from the multi-objective optimization result report and subjected to safety verification, resulting in a safety verification instruction report. Preventive maintenance recommendations are formulated based on the battery health prediction report, resulting in a long-term strategy recommendation report. The safety verification instruction report and the long-term strategy recommendation report are coordinated in terms of short-term and long-term strategies to generate a final execution instruction set. Instructions are distributed and their execution monitored, resulting in a coordinated execution report. The predicted and actual results are compared, and the battery health prediction model is dynamically updated, generating a model calibration report. This invention enables the system to minimize unnecessary reconfiguration switching while ensuring the same equilibrium effect through a battery cell health dynamic prediction module and a multi-objective topology optimization algorithm module, thereby effectively reducing system losses. Furthermore, the invention combines a preventative management strategy planning module with a multi-objective topology optimization algorithm module and a multi-objective decision safety verification module to minimize unnecessary switching caused by excessive pursuit of instantaneous equilibrium. This allows the system to maintain high energy availability while effectively reducing maintenance costs and failure rates, thus improving overall system economy. Finally, the online calibration learning module enables the system to continuously optimize and self-optimize, ensuring maximum control accuracy even under prolonged use. Overall, this invention has significant advantages in terms of strong equilibrium persistence, high multi-performance compatibility, and good battery model adaptive performance. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of an active reconfiguration control system for dynamic balancing of energy storage strings according to the present invention. Figure 2 This is a schematic diagram of an active reconfiguration control method for dynamic balancing of energy storage strings according to the present invention. Detailed Implementation

[0008] 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 only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0009] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” used in the embodiments of this invention are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0010] Depending on the context, the words “if” or “suppose” as used here can be interpreted as “when” or “in response to determination” or “in response to detection.” Similarly, depending on the context, the phrases “if determination” or “if detection (of the stated condition or event)” can be interpreted as “when determination” or “in response to determination” or “when detection (of the stated condition or event)” or “in response to detection (of the stated condition or event).”

[0011] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0012] In practice, the server-side equipment deployed in the active reconfiguration control system for dynamic balancing of energy storage strings may consist of one or more devices. This active reconfiguration control system for dynamic balancing of energy storage strings can be implemented as: a service instance, a virtual machine, or hardware devices. For example, this active reconfiguration control system for dynamic balancing of energy storage strings can be implemented as a service instance deployed on one or more devices in a cloud node. Simply put, this active reconfiguration control system for dynamic balancing of energy storage strings can be understood as software deployed on a cloud node, used to provide active reconfiguration control for dynamic balancing of energy storage strings to each user terminal. Alternatively, this active reconfiguration control system for dynamic balancing of energy storage strings can also be implemented as a virtual machine deployed on one or more devices in a cloud node. This virtual machine contains application software for managing each user terminal. Alternatively, this active reconfiguration control system for dynamic balancing of energy storage strings can also be implemented as a server composed of numerous identical or different types of hardware devices, with one or more hardware devices configured to provide active reconfiguration control for dynamic balancing of energy storage strings to each user terminal.

[0013] In terms of implementation, the active reconfiguration control system for dynamic balancing of energy storage strings and the user terminal are mutually adaptable. That is, if the active reconfiguration control system for dynamic balancing of energy storage strings is an application installed on a cloud service platform, then the user terminal is a client that establishes a communication connection with the application; or if the active reconfiguration control system for dynamic balancing of energy storage strings is implemented as a website, then the user terminal is implemented as a webpage; or if the active reconfiguration control system for dynamic balancing of energy storage strings is implemented as a cloud service platform, then the user terminal is implemented as a mini-program in an instant messaging application.

[0014] like Figure 1The figure shown is a system architecture diagram of an active reconfiguration control system for dynamic balancing of energy storage strings provided in an embodiment of the present invention.

[0015] The active reconfiguration control system for dynamic balancing of energy storage strings described in this invention can be located on a cloud server. In terms of implementation, it can function as one or more service devices, or as an application installed in the cloud (e.g., a mobile service operator's server or server cluster), or it can be developed into a website. Depending on the functions implemented, the active reconfiguration control system for dynamic balancing of energy storage strings may include a data acquisition and feature extraction module, a battery cell health dynamic prediction module, a multi-objective topology optimization algorithm module, a multi-objective decision-making safety verification module, a preventative management strategy planning module, a dynamic coordination command distribution module, and an online calibration learning module. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.

[0016] In this embodiment of the invention, in the active reconfiguration control system for dynamic balancing of energy storage strings, each of the above-mentioned modules can be implemented independently and can call other modules. Here, "calling" can be understood as a module connecting to multiple modules of another type and providing corresponding services to those connected modules. For example, the online calibration learning module can call the same information acquisition module to obtain information collected by that module. Based on the above characteristics, in the active reconfiguration control system for dynamic balancing of energy storage strings provided in this embodiment of the invention, without modifying the program code, the applicable scope of the active reconfiguration control system architecture for dynamic balancing of energy storage strings can be adjusted by adding modules and directly calling them, achieving cluster-based horizontal expansion, so as to achieve the purpose of quickly and flexibly expanding the active reconfiguration control system for dynamic balancing of energy storage strings. In practical applications, the above-mentioned modules can be set in the same device or different devices, or they can be set in virtual devices, such as service instances in a cloud server.

[0017] Example 1 Please see Figure 1 As shown in this embodiment, an active reconfiguration control system for dynamic balancing of energy storage strings includes: The data acquisition feature extraction module is used to send data acquisition instructions to the BMS system based on the acquisition conditions, obtain raw measurement data, and perform preprocessing and feature extraction to obtain a preprocessed feature dataset. Furthermore, the steps of sending data acquisition commands to the BMS system based on the acquisition conditions to obtain raw measurement data, and performing preprocessing and feature extraction include: S1.1: Send data acquisition instructions to the BMS system based on the acquisition conditions to obtain the raw measurement dataset. The data acquisition instructions include high-frequency impedance spectrum measurement instructions and basic data acquisition instructions. It should be explained that the acquisition conditions are manually set and input into the system, for example, the acquisition conditions are preset acquisition cycles; the acquisition data of the high-frequency impedance spectrum measurement command includes the excitation frequency range, excitation amplitude, acquisition timestamp and sampling rate; the acquisition data of the basic data acquisition command is the basic operation data and acquisition timestamp, whereby the basic operation data includes voltage data, current data and temperature data; S1.2: Based on the acquisition timestamp and according to the battery cells, all data items in the original measurement dataset are aligned to obtain the aligned dataset; S1.3: Based on the aligned dataset, the moving average filter is used for processing, and the 3σ method is used for outlier removal. At the same time, linear interpolation is used for padding to obtain the preprocessed dataset. S1.4: Based on the preprocessed dataset, feature extraction is performed to obtain a feature vector set. The features extracted include ohmic internal resistance, charge transfer resistance, diffusion impedance characteristic frequency, and phase angle characteristics. It needs to be explained that, taking the ohmic internal resistance as an example, by substituting into the calculation formula: , obtained the The ohmic resistance of each battery cell ,in, To perform the operation of taking the real part, For frequency The complex impedance below; S1.5: Based on the feature vector set, a multi-dimensional feature vector is constructed for each battery cell to obtain a preprocessed feature dataset; It should be explained that, based on the battery cell, each multidimensional feature vector includes ohmic internal resistance, charge transfer resistance, diffusion impedance characteristic frequency, phase angle characteristic, temperature mean and temperature standard deviation. S1.6: Store the preprocessed feature dataset in the database; The battery cell health dynamic prediction module is used to calculate and predict the health status of each battery cell in real time based on the preprocessed feature dataset, and obtain a battery health prediction report. Furthermore, the steps for real-time calculation and prediction of the health status of each battery cell based on the preprocessed feature dataset include: S2.1: Based on the database, retrieve the preprocessed feature dataset of all battery cells for the past L cycles to obtain the cycle feature sequence; S2.2: Based on the periodic feature sequence, the current health status of the battery is calculated using a pre-trained gradient boosting tree model to obtain a health assessment sequence. Simultaneously, the capacity decay rate is calculated to obtain a capacity decay sequence. The specific formula set for the calculation is as follows: ; The first one was obtained respectively Health assessment value of each battery cell and the Capacity decay rate of individual battery cells ,in, For the first Each battery cell at time point Preprocessed feature data, The set of all learnable parameters for the gradient boosting tree model. This is the cumulative number of loops; S2.3: Input the periodic feature sequence into the battery health prediction model and output the battery health prediction sequence; S2.4: Based on the battery health prediction sequence, the Monte Carlo Dropout method is used to evaluate the prediction uncertainty, and the prediction variance sequence and 95% confidence interval are obtained; S2.5: Package the health assessment sequence, capacity decay sequence, battery health prediction sequence, prediction variance sequence, and 95% confidence interval to obtain the battery health prediction report; S2.6: Output the battery health prediction report to the multi-objective topology optimization algorithm module and the preventive management strategy planning module respectively; The multi-objective topology optimization algorithm module is used to calculate the optimal string connection topology scheme based on the battery health prediction report and obtain a multi-objective optimization result report. Furthermore, the steps for calculating the optimal string connection topology based on the battery health prediction report include: S3.1: Based on the battery health prediction report, define the connection matrix and constraints, where the constraints are that each battery cell must be connected to one and only one string, and the number of batteries in each string is equal; It should be explained that the connection matrix is ​​composed of battery cells and strings; S3.2: Define the objective function based on the battery health prediction report and the connection matrix to obtain a set of objective functions. The objective functions include minimizing the difference in battery health prediction values ​​within the group, minimizing the switching cost, and maximizing the consistency of battery health prediction values ​​within the group. S3.3: Use the NSGA-II algorithm to randomly generate U connection matrices that satisfy the constraints, and perform iterative optimization. When the target number of iterations is reached, obtain the Pareto front solution set. S3.4: For each Pareto solution, calculate the performance index and obtain the topology performance evaluation report. The performance index includes the objective function, the efficiency improvement evaluation value, and the number of switching operations. S3.5: Package the Pareto front solution set and topological performance evaluation report to obtain the multi-objective optimization result report; S3.6: Output the multi-objective optimization result report to the multi-objective decision security verification module; The multi-objective decision security verification module is used to select the final execution plan from the multi-objective optimization result report, perform security verification, and obtain a security verification instruction report. Furthermore, the steps for selecting the final implementation plan from the multi-objective optimization results report and performing security verification include: S4.1: Based on the long-term strategy guidance vector and combined with the current working conditions, the weights are adjusted to obtain the comprehensive weights; It should be explained that the current operating condition refers to, for example, high power demand or high temperature environment; S4.2: Based on the Pareto front solution set in the multi-objective optimization result report, calculate the weighted score for each Pareto solution and select the solution with the highest score as the candidate solution; S4.3: Perform simulated safety verification based on candidate solutions. If the safety verification result is passed, proceed to step S4.4. If the safety verification result is failed, terminate the operation and generate a safety verification report to be output to the staff receiving end. The simulated safety verification includes electrical safety verification, thermal safety simulation and switching capability verification. S4.4: Compare the execution plan of the current connection with the candidate plan output in step S4.3 to generate a reconstruction instruction sequence; It should be explained that the refactoring instruction sequence includes the toggle switch ID, source group, grouping, and execution time; S4.5: Based on steps S4.1 to S4.4, generate a security verification report; It should be explained that the safety verification report includes, but is not limited to, safety verification results, voltage compliance, temperature prediction values, and switch load rate; S4.6: Package and reconstruct the instruction sequence and security verification report to obtain the security verification instruction report; S4.7: Output the security verification instruction report to the dynamic coordination instruction distribution module; The preventive management strategy planning module is used to formulate preventive maintenance recommendations based on the battery health prediction report and obtain a long-term strategy recommendation report. Furthermore, the steps for developing preventative maintenance recommendations based on battery health prediction reports include: S5.1: Calculate the degradation rate of each battery cell based on the battery health prediction report to obtain the degradation rate sequence; S5.2: Based on the decay rate sequence, the clustering distance is calculated using the Ward method, and the battery cells are divided into C homogeneous groups to obtain the battery cluster set; S5.3: Based on battery clusters and according to the degradation threshold, battery clusters with degradation rates less than the degradation threshold are classified as fast degradation groups, and battery clusters with degradation rates greater than or equal to the degradation threshold are classified as stable groups. Corresponding strategies are formulated according to the differences between fast degradation groups and stable groups to obtain a long-term strategy report. It should be explained that the degradation threshold is manually set and input into the system; the corresponding strategy means, for example, for the rapid degradation group, the long-term strategy is to reduce the charge and discharge rate, and for the stable group, the long-term strategy is to extend the maintenance cycle. S5.4: Generate long-term strategy guidance vectors based on the current system stage, which includes the initial, middle and final stages; It should be explained that, taking the current system stage as the initial stage as an example, the long-term strategy guidance vector is a set of objective functions of 0.6, 0.2 and 0.2 that emphasize equilibrium, which correspond to the three objective functions in step S3.2 respectively; S5.5: Based on steps S5.1 to S5.4, generate maintenance recommendations and obtain a preventive maintenance recommendation report; It should be explained that the maintenance recommendations include, but are not limited to, a list of battery cells that require special attention, a recommended maintenance schedule, and the distribution of expected remaining lifespan. S5.6: Package the long-term strategy guidance vector and preventive maintenance recommendation report to obtain the long-term strategy recommendation report; S5.7: Output the long-term strategy recommendation report to the multi-objective decision security verification module; The dynamic coordination instruction distribution module is used to coordinate short-term and long-term strategies for security verification instruction reports and long-term strategy suggestion reports, generate the final execution instruction set, and perform instruction distribution and execution monitoring to obtain a coordination execution report. Furthermore, the steps of coordinating short-term and long-term strategies in the security verification instruction report and long-term strategy recommendation report to generate the final execution instruction set, and distributing and monitoring the execution of the instructions include: S6.1: Calculate the imbalance of the current system based on the security verification instruction report to obtain the reconstruction evaluation value. When the reconstruction evaluation value is less than the reconstruction threshold, skip the reconstruction. When the number of skips is greater than the maximum allowed number of skips, force the reconstruction to be triggered. It should be explained that the reconstruction threshold was manually set and input into the system; S6.2: Based on the long-term strategy guidance vector and combined with the current working conditions, the optimal reconfiguration time is calculated to obtain the optimal reconfiguration time point; It should be explained that the calculation in step S6.2 refers to minimizing the weighted sum of the predicted imbalance and the switching cost; S6.3: Verify whether the reconstructed instruction sequence is feasible at the current time. If the verification result is not feasible, return to the multi-objective topology optimization algorithm module for recalculation. If the verification result is feasible, package the reconstructed instruction sequence and the execution time window to obtain the final execution instruction set, and proceed to step S6.4. S6.4: Distribute the final execution instruction set to the switch execution system, monitor the execution process, and obtain a coordinated execution report; It should be explained that the coordination and execution report includes the instruction distribution time, receipt status, execution progress feedback, and actual completion time; S6.5: Output the coordinated execution report to the online calibration learning module; The online calibration learning module is used to compare the predicted and actual results, dynamically update the battery health prediction model, and generate a model calibration report. Furthermore, the steps of comparing the predictions with the actual results and dynamically updating the battery health prediction model include: S7.1: Obtain the battery health prediction sequence based on the battery health prediction report, and calculate the prediction error value sequence by comparing it with the actual battery health value at the same time. It should be explained that the calculation in step S7.1 refers to the absolute value obtained by subtracting the actual battery health value from the predicted battery health value; S7.2: Calculate the overall prediction accuracy index based on the prediction error value sequence to obtain a prediction accuracy report. The prediction accuracy index includes the mean absolute error, root mean square error, and maximum error. S7.3: When the mean absolute error is greater than the mean absolute error threshold, generate a model calibration signal and proceed to step S7.4; It should be explained that the mean absolute error threshold was manually set and entered into the system; S7.4: When a model calibration signal appears, construct a training dataset and adjust the battery health prediction model parameters to obtain a model calibration report; S7.5: Output the model calibration report to the battery cell health dynamic prediction module; The beneficial effects of this embodiment are as follows: Data acquisition commands are sent to the BMS system based on acquisition conditions to obtain raw measurement data. This data is then preprocessed and features extracted to obtain a preprocessed feature dataset. Based on this dataset, the health status of each battery cell is calculated and predicted in real time, resulting in a battery health prediction report. The optimal string connection topology is calculated based on the battery health prediction report, resulting in a multi-objective optimization result report. The final execution plan is selected from the multi-objective optimization result report and safety verification is performed, resulting in a safety verification instruction report. Preventive maintenance recommendations are formulated based on the battery health prediction report, resulting in a long-term strategy recommendation report. The safety verification instruction report and the long-term strategy recommendation report are coordinated in terms of both short-term and long-term strategies to generate a final execution instruction set. Instructions are distributed and execution is monitored, resulting in a coordinated execution report. The predicted and actual results are compared, and the battery health prediction model is dynamically updated, generating a model calibration report. The report enables the system to minimize unnecessary reconfiguration switching while ensuring the same equilibrium effect through the battery cell health dynamic prediction module and the multi-objective topology optimization algorithm module, thereby effectively reducing system losses. Furthermore, this embodiment combines a preventative management strategy planning module with the multi-objective topology optimization algorithm module and the multi-objective decision safety verification module to avoid unnecessary switching due to an excessive pursuit of instantaneous balance. This allows the system to effectively reduce operation and maintenance costs and failure rates while ensuring high energy availability, thus improving the overall economic efficiency of the system. Finally, the online calibration learning module enables the system to continuously optimize and self-optimize, maintaining control accuracy to the greatest extent even under long-term use. Overall, this embodiment has significant advantages such as strong equilibrium persistence, high performance balance, and good battery model adaptive effect. Example 2 Please see Figure 2 As shown, the parts not described in detail in this embodiment are described in Embodiment 1. An active reconfiguration control method for dynamic balancing of energy storage strings is provided. The method includes: S1: sending a data acquisition command to the BMS system based on the acquisition conditions to obtain the raw measurement data, and performing preprocessing and feature extraction to obtain a preprocessed feature dataset. S2: Calculate and predict the health status of each battery cell in real time based on the preprocessed feature dataset to obtain a battery health prediction report; S3: Calculate the optimal string connection topology based on the battery health prediction report and obtain a multi-objective optimization result report; S4: Select the final execution plan from the multi-objective optimization result report, perform security verification, and obtain a security verification instruction report; S5: Based on the battery health prediction report, preventive maintenance recommendations are formulated to obtain a long-term strategy recommendation report; S6: Coordinate short-term and long-term strategies for security verification instruction reports and long-term strategy recommendation reports, generate the final execution instruction set, distribute and monitor the execution of instructions, and obtain a coordinated execution report; S7: Compare the predicted and actual results, dynamically update the battery health prediction model, and generate a model calibration report.

[0018] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the present invention.

Claims

1. An active reconfiguration control system for dynamic equalization of energy storage strings, characterized by, The system comprises a battery cell health dynamic prediction module, a multi-objective topology optimization algorithm module, a multi-objective decision safety verification module, a preventive management strategy planning module, and an online calibration learning module, wherein: The battery cell health dynamic prediction module is configured to calculate and predict the health state of each battery cell in real time based on the preprocessed feature dataset, and obtain a battery health prediction report; The multi-objective topology optimization algorithm module is configured to calculate an optimal group string connection topology scheme based on the battery health prediction report, and obtain a multi-objective optimization result report; The multi-objective decision safety verification module is configured to select a final execution scheme from the multi-objective optimization result report and perform safety verification, and obtain a safety verification instruction report; The preventive management strategy planning module is configured to formulate a preventive maintenance suggestion based on the battery health prediction report, and obtain a long-term strategy suggestion report; The online calibration learning module is configured to compare the prediction result with the actual result, and dynamically update the battery health prediction model, and generate a model calibration report.

2. The active reconfiguration control system for dynamic equalization of energy storage strings of claim 1, wherein, The system further comprises a data acquisition feature extraction module and a dynamic coordination instruction distribution module, wherein: The data acquisition feature extraction module is configured to send a data acquisition instruction to the BMS system based on the acquisition condition, obtain raw measurement data, and perform preprocessing and feature extraction to obtain a preprocessed feature dataset; The dynamic coordination instruction distribution module is configured to coordinate long-term and short-term strategies based on the safety verification instruction report and the long-term strategy suggestion report, generate a final execution instruction set, and perform instruction distribution and execution monitoring to obtain a coordination execution report.

3. The active reconfiguration control system for dynamic equalization of energy storage strings of claim 2, wherein, The steps of sending a data acquisition instruction to the BMS system based on the acquisition condition, obtaining raw measurement data, and performing preprocessing and feature extraction include: S1.1: Send a data acquisition instruction to the BMS system based on the acquisition condition, and obtain a raw measurement dataset, wherein the data acquisition instruction includes high-frequency impedance spectrum measurement instructions and basic data acquisition instructions; S1.2: Align all data items in the raw measurement dataset based on the acquisition timestamp and according to the battery cell, and obtain an aligned dataset; S1.3: Process the aligned dataset using sliding average filtering, remove outliers using the 3σ method, and fill in missing values using linear interpolation, and obtain a preprocessed dataset; S1.4: Extract features from the preprocessed dataset to obtain a feature vector set, wherein the features include ohmic resistance, charge transfer resistance, diffusion impedance characteristic frequency, and phase angle characteristic; S1.5: Construct a multi-dimensional feature vector for each battery cell based on the feature vector set, and obtain a preprocessed feature dataset; S1.6: Store the preprocessed feature dataset in a database.

4. The active reconfiguration control system for dynamic equalization of energy storage strings of claim 3, wherein, The steps of calculating and predicting the health state of each battery cell in real time based on the preprocessed feature dataset include: S2.1: Retrieve the preprocessed feature dataset of all battery cells for L historical periods from the database to obtain a period feature sequence; S2.2: Based on the periodic feature sequence, the current state of health of the battery is calculated using a pre-trained gradient boosting tree model to obtain a health assessment sequence, and the capacity decay rate is calculated at the same time to obtain a capacity decay sequence; S2.3: Input the periodic feature sequence into the battery health prediction model to output the battery health prediction sequence; S2.4: Based on the battery health prediction sequence, the prediction uncertainty is evaluated using the Monte Carlo Dropout method to obtain the prediction variance sequence and 95% confidence interval; S2.5: Package the health assessment sequence, capacity decay sequence, battery health prediction sequence, prediction variance sequence and 95% confidence interval to obtain the battery health prediction report; S2.6: Output the battery health prediction report to the multi-objective topology optimization algorithm module and the preventive management strategy planning module respectively.

5. The active reconfiguration control system for dynamic equalization of energy storage strings of claim 4, wherein, The steps for calculating the optimal group string connection topology scheme based on the battery health prediction report include: S3.1: Based on the battery health prediction report, define the connection matrix and the constraint condition, wherein the constraint condition is that each battery unit must and can only be connected to one group string, and the number of batteries in each group string is equal; S3.2: Based on the battery health prediction report and the connection matrix, define the objective function to obtain the objective function group, which includes minimizing the difference of the battery health prediction value within the group, minimizing the switching cost of the switch, and maximizing the consistency of the battery health prediction value within the group; S3.3: Randomly generate U connection matrices that satisfy the constraint condition using the NSGA-II algorithm, and perform iterative optimization. When the target iteration number is reached, the Pareto frontier solution set is obtained; S3.4: For each Pareto solution, calculate the performance index to obtain a topology performance evaluation report, and the performance index includes the objective function, efficiency improvement evaluation value and switch number; S3.5: Package the Pareto frontier solution set and the topology performance evaluation report to obtain a multi-objective optimization result report; S3.6: Output the multi-objective optimization result report to the multi-objective decision safety verification module.

6. The active reconfiguration control system for dynamic equalization of energy storage strings of claim 1, wherein, The steps for selecting the final execution scheme from the multi-objective optimization result report and performing safety verification include: S4.1: Based on the long-term strategy guide vector and combined with the current working condition to adjust the weight to obtain the comprehensive weight; S4.2: Based on the Pareto frontier solution set in the multi-objective optimization result report, calculate the weighted score for each Pareto solution, and select the solution with the highest score as the candidate scheme; S4.3: Based on the candidate scheme, perform simulation safety verification. When the safety verification result is passed, go to step S4.

4. When the safety verification result is not passed, terminate the operation and generate a safety verification report output to the staff receiving end. The simulation safety verification includes electrical safety verification, thermal safety simulation and switch capacity verification; S4.4: Compare the current connected execution scheme with the candidate scheme output in step S4.3 to generate a reconstruction instruction sequence; S4.5: Based on steps S4.1 to S4.4, generate a safety verification report; S4.6: Package the reconstruction instruction sequence and the safety verification report to obtain a safety verification instruction report; S4.7: Output the safety verification instruction report to the dynamic coordination instruction distribution module.

7. The active reconfiguration control system for dynamic equalization of energy storage strings of claim 4, wherein, The steps of formulating the preventive maintenance suggestion based on the battery health prediction report include: S5.1: Calculate the degradation rate of each battery unit based on the battery health prediction report to obtain a degradation rate sequence; S5.2: Calculate the clustering distance based on the degradation rate sequence and use the Ward method to divide the battery units into C homogeneous groups to obtain a battery clustering group set; S5.3: Divide the battery clustering group set based on the degradation threshold, divide the battery clustering group with a degradation rate less than the degradation threshold into a fast degradation group, and divide the battery clustering group with a degradation rate greater than or equal to the degradation threshold into a stable group, and formulate corresponding strategies according to the difference between the fast degradation group and the stable group to obtain a long-term strategy report; S5.4: Generate a long-term strategy guidance vector based on the current system stage, which includes the initial stage, the middle stage, and the final stage; S5.5: Generate a maintenance suggestion based on steps S5.1 to S5.4 to obtain a preventive maintenance suggestion report; S5.6: Package the long-term strategy guidance vector and the preventive maintenance suggestion report to obtain a long-term strategy suggestion report; S5.7: Output the long-term strategy suggestion report to the multi-objective decision safety verification module.

8. The active reconfiguration control system for dynamic equalization of energy storage strings of claim 6, wherein, The steps of coordinating the long-term and short-term strategies based on the safety verification instruction report and the long-term strategy suggestion report, generating a final execution instruction set, and performing instruction distribution and execution monitoring include: S6.1: Calculate the imbalance degree of the current system based on the safety verification instruction report to obtain a reconstruction evaluation value, when the reconstruction evaluation value is less than the reconstruction threshold, skip this reconstruction, when the number of skips is greater than the maximum allowed number of skips, force trigger the reconstruction; S6.2: Calculate the optimal reconstruction time based on the long-term strategy guidance vector and the current working condition to obtain an optimal reconstruction time point; S6.3: Verify whether the reconstruction instruction sequence is feasible at the current time, when the verification result is not feasible, return to the multi-objective topology optimization algorithm module to recalculate, when the verification result is feasible, package the reconstruction instruction sequence and the execution time window to obtain a final execution instruction set, and enter step S6.4; S6.4: Distribute the final execution instruction set to the switch execution system and monitor the execution process to obtain a coordinated execution report; S6.5: Output the coordinated execution report to the online calibration learning module.

9. The active reconfiguration control system for dynamic equalization of energy storage strings of claim 4, wherein, The steps of comparing the prediction and actual results and dynamically updating the battery health prediction model include: S7.1: Obtain the battery health prediction sequence based on the battery health prediction report and calculate the actual battery health value at the same time to obtain a prediction error value sequence; S7.2: Calculate the overall prediction accuracy index based on the prediction error value sequence to obtain a prediction accuracy report, wherein the prediction accuracy index includes the mean absolute error, the root mean square error, and the maximum error; S7.3: When the mean absolute error is greater than the mean absolute error threshold, generate a model calibration signal and enter step S7.4; S7.4: When the model calibration signal appears, construct a training data set and adjust the battery health prediction model parameters to obtain a model calibration report; S7.5: output the model calibration report to the battery cell health dynamic prediction module.

10. An active reconfiguration control method for dynamic equalization of energy storage strings, implemented by the active reconfiguration control system for dynamic equalization of energy storage strings according to any one of claims 1-9, characterized in that, The following working steps are included: S1: send data acquisition instructions to the BMS system based on the acquisition conditions, obtain the original measurement data, and perform preprocessing and feature extraction to obtain a preprocessed feature dataset; S2: based on the preprocessed feature dataset, calculate and predict the health status of each battery cell in real time to obtain a battery health prediction report; S3: based on the battery health prediction report, calculate the optimal group string connection topology scheme to obtain a multi-objective optimization result report; S4: select the final execution scheme from the multi-objective optimization result report and perform safety verification to obtain a safety verification instruction report; S5: based on the battery health prediction report, formulate preventive maintenance recommendations to obtain a long-term strategy recommendation report; S6: coordinate the safety verification instruction report and the long-term strategy recommendation report for long-term and short-term strategy coordination, generate a final execution instruction set, and perform instruction distribution and execution monitoring to obtain a coordinated execution report; S7: compare the predicted and actual results, and dynamically update the battery health prediction model to generate a model calibration report.

Images