Battery energy storage system state evaluation system and method

By combining rough set theory and GA-BP neural network, the problem of insufficient accuracy in battery energy storage system status evaluation is solved, and high-precision SOC status evaluation is achieved, which improves the safety and life of the battery management system.

CN118885779APending Publication Date: 2024-11-01JIYUAN POWER SUPPLY COMPANY OF STATE GRID HENAN ELECTRIC POWER
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202410795335.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-06-19
Publication Date
2024-11-01

AI Technical Summary

Technical Problem

The accuracy of the existing battery energy storage system state evaluation method is poor, especially because the battery is a complex nonlinear system, it is difficult to establish an accurate mathematical model, resulting in insufficient SOC estimation accuracy.

Method used

Using a combination of data mining based on rough set theory and GA-BP neural network, the charge and discharge test data of the battery energy storage system are used to reduce attributes and build neural network models to achieve high-precision evaluation of SOC state.

Benefits of technology

It improves the prediction accuracy of SOC status evaluation in battery energy storage system, can more accurately analyze the main factors affecting SOC, and enhances the safety and life of the battery management system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118885779A_ABST
    Figure CN118885779A_ABST
Patent Text Reader

Abstract

The invention relates to a battery energy storage system state evaluation system, which comprises an energy storage battery management system, and the energy storage battery management system comprises an energy storage battery system, an MATLAB simulation module, a core program set module, a MySQL database and a WEB management demonstration platform. The MATLAB simulation module simulates a real energy storage battery system by using a battery model, can output state parameters of a battery, can receive input and changes the state parameters according to the input; the core program set module is responsible for core calculation of the battery management system; the MySQL database is used for storing battery parameters and user information which are updated in real time, and state parameters and auxiliary data of the core program set module; the Web management demonstration platform is used for providing a battery management system entrance, so that the operation state of the system can be known in real time, and a corresponding instruction is sent out; the method has the advantages that data mining based on the rough set theory is achieved, state evaluation is achieved through the neural network, and prediction precision is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of battery energy storage system status assessment, and in particular relates to a battery energy storage system status assessment system and method. Background Art

[0002] As the country increases its investment in the construction of the next generation of smart grids, research on related technologies has gradually become popular. Among them, energy storage technology, especially battery energy storage technology, has attracted attention because of its ability to store electricity generated by unstable renewable energy (such as wind energy and solar energy). However, the operation of battery energy storage systems requires battery management to ensure battery safety and life; SOC estimation methods are roughly divided into two types: traditional estimation methods and advanced estimation methods. Traditional estimation methods include current integration method, discharge test method, open circuit voltage method, load voltage method, electrochemical impedance spectroscopy, internal resistance method, etc. This type of method realizes SOC estimation by physically measuring parameters such as current and voltage. The method is simple. But the accuracy is poor; the advanced estimation method mainly applies some intelligent methods (such as neural networks, support vector machines) to model the battery system, integrates various factors affecting SOC into the battery model, and uses a large number of experiments to identify the model system and estimate the parameters, and obtain the relationship between certain battery parameters and SOC, and then estimate SOC. Since the battery is a complex nonlinear system, it is difficult to establish an accurate mathematical model for its charging and discharging process; therefore, it is very necessary to provide a battery energy storage system state evaluation system and method based on rough set theory data mining, using neural networks to achieve state evaluation, and improve prediction accuracy. Summary of the invention

[0003] The purpose of the present invention is to overcome the deficiencies of the prior art and to provide a battery energy storage system state assessment system and method based on data mining based on rough set theory, using neural networks to achieve state assessment and improve prediction accuracy.

[0004] The object of the present invention is achieved as follows: a battery energy storage system state evaluation system, which includes an energy storage battery management system, the energy storage battery management system includes an energy storage battery system, a MATLAB simulation module, a core assembly module, a MySQL database, and a WEB management demonstration platform; the MATLAB simulation module uses a battery model to simulate a real energy storage battery system, can output battery state parameters, and can accept input and change state parameters according to the input; the core assembly module is responsible for the core calculation of the battery management system, including; the MySQL database is used to store real-time updated battery parameters, user information, and state parameters and auxiliary data of the core assembly module; the Web management demonstration platform is used to provide a battery management system entrance, so as to facilitate real-time understanding of the system's operating status and issue corresponding instructions.

[0005] A battery energy storage system state assessment method adopts a battery energy storage system SOC state assessment based on data mining technology and is applied to the above-mentioned battery energy storage system state assessment system. The method comprises the following steps:

[0006] Step 1: Battery energy storage system charging and discharging test process;

[0007] Step 2: Attribute reduction method based on rough set theory;

[0008] Step 3: GA-BP neural network battery SOC state evaluation method based on rough set attribute reduction.

[0009] The battery energy storage system charge and discharge test process in step 1 includes the following steps:

[0010] Step 1.1: Battery SOC: SOC represents the remaining battery power or remaining operating time in the current charging cycle. The ratio of the remaining capacity during battery operation to the rated capacity of the battery is usually called the battery SOC: In the formula, Q C is the remaining capacity of the battery; Q I is the rated capacity of the battery;

[0011] Step 1.2: Battery variable current intermittent charging and discharging process: The collected data include the battery current I, voltage U, internal resistance R, power P, cycle power change W1, cycle charge W2, cycle discharge W3, battery temperature T and battery remaining capacity Q at each moment. The SOC value of the entire moment is calculated using the remaining capacity value Q at each moment according to formula (1).

[0012] The attribute simplification method based on rough set theory in step 2 is specifically as follows: starting from the empty set, calculating the attribute importance of the remaining attributes each time, selecting attributes with large attribute importance and incorporating them into the simplified set until the simplification conditions are met, ensuring that important attributes are added to the simplification first, so as to facilitate rapid finding of the simplification.

[0013] The reduced set in step 2 is: according to the importance of the conditional attribute, the reduced set can be obtained by judging whether the attribute importance is greater than zero in the neighborhood rough set reduction.

[0014] The GA-BP neural network battery SOC state evaluation method based on rough set attribute reduction in step 3 comprises the following steps:

[0015] Step 3.1: Construct a SOC prediction model based on BP neural network;

[0016] Step 3.2: GA-BP neural network;

[0017] Step 3.3: Battery SOC state evaluation method based on RS-GA-BP neural network.

[0018] The construction of the SOC prediction model based on the BP neural network in step 3.1 is specifically as follows: the working voltage U, current I and temperature T of the battery energy storage system at a certain moment are the main factors affecting the SOC, so these three factors are used as the input vector [X1, X2, X3] of the model, that is, the number of input nodes of the neural network is 3, and the battery SOC is used as the output vector [Y] of the model, that is, the number of output nodes is 1; 12 nodes are used in the hidden layer to accurately describe the relationship between the battery voltage, current, temperature and the battery SOC.

[0019] The GA-BP neural network in step 3.2 includes the following steps:

[0020] Step 3.21: Generate the initial population step;

[0021] Step 3.22: Determine the relevant parameters and termination conditions of the genetic algorithm, initialize the genetic algorithm and perform calculations;

[0022] Step 3.23: Evaluate the calculation results, decode the BP network and calculate the fitness value of each chromosome;

[0023] Step 3.24: Determine whether the genetic algorithm has reached the termination condition. If so, the calculation ends and the algorithm jumps to step 3.26 for execution. If not, the algorithm continues to execute downward.

[0024] Step 3.25: Perform selection, crossover, and mutation genetic algorithm operations to generate a new generation of chromosomes, and then go to step 3.23 for a loop operation;

[0025] Step 3.26: Use the BP network to locally fine-tune the approximate optimal value searched by the genetic algorithm to improve the accuracy of the solution until the accuracy requirements are met and proceed to the next step;

[0026] Step 3.27: Save the BP network model calculated by the sample and input the test sample for comparative analysis;

[0027] Step 3.28: Output the results.

[0028] The battery SOC state evaluation method based on RS-GA-BP neural network in step 3.3 comprises the following steps:

[0029] Step 3.31: Use rough sets as a precondition system;

[0030] Step 3.32: Preprocess the samples before SOC state prediction by removing redundant attributes and identical samples through attribute simplification;

[0031] Step 3.33: Then send the data preprocessed by neighborhood rough set into GA-BP neural network for prediction.

[0032] Beneficial effects of the present invention: The present invention is a battery energy storage system state evaluation system and method, which analyzes the main factors affecting the SOC operating state of the battery energy storage system, and adopts a battery state of charge SOC evaluation method based on a GA-BP neural network model based on data mining. In use, firstly, the measured battery energy storage system operating state data is mined based on a data mining method based on rough set theory, data mining of factors affecting SOC is realized, and important parameters affecting SOC are extracted; finally, the mining results are combined with a GA-BP neural network algorithm, and the GA-BP neural network is used to realize the battery state of charge SOC state evaluation, with high prediction accuracy; the present invention has the advantages of data mining based on rough set theory, using neural network to realize state evaluation, and improving prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a system block diagram of the present invention.

[0034] Figure 2 It is the SOC curve diagram of the variable current intermittent charge and discharge test of the present invention.

[0035] Figure 3 It is a current and voltage curve diagram of the variable current intermittent charging and discharging process of the present invention.

[0036] Figure 4 This is a flow chart of the rough set attribute reduction algorithm of the present invention.

[0037] Figure 5 It is a schematic diagram of the SOC prediction model based on BP neural network of the present invention.

[0038] Figure 6 It is a flow chart of the GA-BP network prediction model of the present invention.

[0039] Figure 7 This is a flow chart of the RS-GA-BP neural network battery SOC state evaluation of the present invention.

[0040] Figure 8 This is the flow chart of the BP neural network optimized by the MEA algorithm of the present invention. DETAILED DESCRIPTION

[0041] The present invention will be further described below in conjunction with the accompanying drawings.

[0042] Example 1

[0043] like Figure 1As shown, a battery energy storage system state assessment system includes an energy storage battery management system, which includes an energy storage battery system, a MATLAB simulation module, a core assembly module, a MySQL database, and a WEB management demonstration platform; the MATLAB simulation module uses a battery model to simulate a real energy storage battery system, can output battery state parameters, and can accept input and change state parameters according to the input; the core assembly module is responsible for the core calculation of the battery management system, including; the MySQL database is used to store real-time updated battery parameters, user information, and state parameters and auxiliary data of the core assembly module; the Web management demonstration platform is used to provide an entrance to the battery management system, so as to facilitate real-time understanding of the system's operating status and issue corresponding instructions.

[0044] In this embodiment, the energy storage battery management system in the system of the present invention realizes fine-grained and high-real-time charge and discharge control of the battery through software simulation of an energy storage system composed of a certain number of battery cells connected in series, and can provide support for battery balancing and wind-storage complementarity; after the energy storage battery system is introduced into the system, since the difference in parameters of the battery cells will gradually increase with use, thereby seriously affecting the cycle life of the entire energy storage system, it is necessary to manage the battery to ensure the consistency of the battery. The battery management system of the present invention provides a control logic that actively keeps the battery's charge level SOC consistent, that is, battery balancing. The system architecture is as follows Figure 1 shown.

[0045] The present invention is a battery energy storage system state assessment system. In use, the present invention introduces computer and Internet technology into a traditional battery management system based on analog circuits and microcontrollers, and proposes a battery management system in which a computer performs fine-grained monitoring and control of battery cells, thereby bringing higher safety and longer life to the battery system, reducing the complexity and cost of battery-side hardware, and introducing a web-based management demonstration platform into the system to improve the convenience of the system. The energy storage system composed of a plurality of battery cells is simulated using a MATLAB simulation module to achieve monitoring and control of the simulated battery system, and the voltage, current, and SOC parameters of the battery cells can be output in real time, and independent charge and discharge control of the battery cells can be performed according to the input, thereby achieving simulation and demonstration of two main applications of wind-storage complementarity and battery balancing, and each module of the system can coordinately and uniformly achieve the overall function of the system; the present invention has the advantages of data mining based on rough set theory, using neural networks to achieve state assessment, and improving prediction accuracy.

[0046] Example 2

[0047] like Figure 2-8As shown, a battery energy storage system state assessment method adopts a battery energy storage system SOC state assessment based on data mining technology and is applied to the above-mentioned battery energy storage system state assessment system. The method includes the following steps:

[0048] Step 1: Battery energy storage system charging and discharging test process;

[0049] In this embodiment, the battery energy storage system charge and discharge test process includes the following steps:

[0050] Step 1.1: Battery SOC: SOC represents the remaining battery power or remaining operating time in the current charging cycle. The ratio of the remaining capacity during battery operation to the rated capacity of the battery is usually called the battery SOC: In the formula, Q C is the remaining capacity of the battery; Q I is the rated capacity of the battery;

[0051] Step 1.2: Battery variable current intermittent charge and discharge process: A variable current intermittent charge and discharge process test of the all-vanadium liquid flow battery was carried out using a power supply test system in a substation. The SOC curve of the variable current intermittent charge and discharge test is as follows: Figure 2 As shown in the figure, the whole operation time is 5.66h; the current and voltage change curves of the variable current intermittent charging and discharging process are as follows Figure 3 As shown, a relatively small charging current (generally near the rated current) is selected in the first stage, and charging is stopped after a period of time; during the period of stopping charging, the battery voltage drops sharply, and the polarization effect gradually decreases. After stopping charging for a period of time, the battery is discharged in a constant current mode, and the charging is stopped for a period of time when the discharge voltage reaches the cut-off voltage, and the charging current is increased in a certain way and continued to charge for a period of time, and then the charging is stopped again, and the battery is discharged in a constant current mode, and the charging is stopped for a period of time when the discharge voltage reaches the cut-off voltage, and this is repeated several times to increase the charging current until the battery is fully charged, and the maximum charging current is used in the last stage to end the entire charging process;

[0052] The test parameters that need to be determined for the variable current intermittent charge and discharge test include the initial charging current, charging time, conversion voltage, charging stop time, current conversion method, etc.; in the test, the initial charging current is 10A, the final constant current charging current is 180A, the intermittent charging conversion voltage is set to 35V, the charging stop time is constant at 120s, the charging current changes according to the law of increasing arithmetic progression, and the tolerance is 10A. The discharge process test of this system is carried out by discharging with a discharge instrument, and the discharge current is constant at 10A;

[0053] The 20387 sets of data collected during the variable current intermittent charging and discharging process are used for analysis. The data include the battery current I, voltage U, internal resistance R, power P, cycle power change W1, cycle charge W2, cycle discharge W3, battery temperature T and battery remaining capacity Q at each moment. The remaining capacity Q at each moment is used to calculate the SOC value of the entire moment according to formula (1).

[0054] Step 2: Attribute reduction method based on rough set theory;

[0055] In this embodiment, the attribute reduction method based on the rough set theory is specifically as follows: starting from the empty set, calculating the attribute importance of the remaining attributes each time, selecting attributes with large attribute importance and incorporating them into the reduced set until the reduction conditions are met, ensuring that important attributes are added to the reduction first, so as to facilitate rapid finding of the reduction. The rough set attribute reduction algorithm process is as follows: Figure 4 As shown;

[0056] The reduced set in step 2 is: according to the importance of the conditional attribute, the reduced set can be obtained by judging whether the attribute importance is greater than zero in the neighborhood rough set reduction.

[0057] Step 3: GA-BP neural network battery SOC state evaluation method based on rough set attribute reduction.

[0058] In this embodiment, the GA-BP neural network battery SOC state evaluation method based on rough set attribute reduction includes the following steps:

[0059] Step 3.1: Construct a SOC prediction model based on BP neural network, specifically: use a three-layer BP neural network to predict the battery energy storage SOC, such as Figure 5 As shown in the figure, the working voltage U, current I and temperature T of the battery energy storage system at a certain moment are the main factors affecting SOC. Therefore, these three factors are used as the input vector [X1, X2, X3] of the model, that is, the number of input nodes of the neural network is 3, and the battery SOC is used as the output vector [Y] of the model, that is, the number of output nodes is 1; 12 nodes are used in the hidden layer to accurately describe the relationship between battery voltage, current, temperature and battery SOC;

[0060] Step 3.2: GA-BP neural network, specifically: GA-BP genetic neural network first uses genetic algorithm to perform global search, optimizes the weights and thresholds of BP network, and then assigns the neural network weights and threshold vectors to BP network, and uses the local search ability of neural network to obtain the approximate optimal value of the network, ensuring that BP network converges to the global optimum and obtains accurate optimization results, thereby avoiding the shortcomings of BP network such as slow convergence speed, easy to fall into the same local extreme value and poor stability. The GA-BP network prediction model process is as follows Figure 6As shown; including the following steps:

[0061] Step 3.21: Generate the initial population step;

[0062] Step 3.22: Determine the relevant parameters and termination conditions of the genetic algorithm, initialize the genetic algorithm and perform calculations;

[0063] Step 3.23: Evaluate the calculation results, decode the BP network and calculate the fitness value of each chromosome;

[0064] Step 3.24: Determine whether the genetic algorithm has reached the termination condition. If so, the calculation ends and the algorithm jumps to step 3.26 for execution. If not, the algorithm continues to execute downward.

[0065] Step 3.25: Perform selection, crossover, and mutation genetic algorithm operations to generate a new generation of chromosomes, and then go to step 3.23 for a loop operation;

[0066] Step 3.26: Use the BP network to locally fine-tune the approximate optimal value searched by the genetic algorithm to improve the accuracy of the solution until the accuracy requirements are met and proceed to the next step;

[0067] Step 3.27: Save the BP network model calculated by the sample and input the test sample for comparative analysis;

[0068] Step 3.28: Output the results.

[0069] Step 3.3: The battery SOC state evaluation method based on RS-GA-BP neural network includes the following steps:

[0070] Step 3.31: Use rough sets as a precondition system;

[0071] Step 3.32: Preprocess the samples before SOC state prediction by removing redundant attributes and identical samples through attribute simplification;

[0072] Step 3.33: Send the data after neighborhood rough set preprocessing into the GA-BP neural network for prediction, simplify the structure of the neural network, speed up the network processing speed, and improve the prediction accuracy. The RS-GA-BP neural network battery SOC state evaluation flow chart is as follows: Figure 7 shown.

[0073] In this embodiment, the GA-BP neural network in step 3.2 and the battery SOC state evaluation method based on the RS-GA-BP neural network in step 3.3 can be replaced by: MEA-BP neural network SOC prediction model, which specifically includes the following steps:

[0074] ① Use the initialization population generation function to generate a population of size Nsize The population size; the number of subpopulations generated by the subpopulation generation function is N bsize The number of winner subpopulations is N tsize temporary subpopulation;

[0075] ② Convergence operation of sub-populations: The process of individuals competing to become winners within the sub-population is called convergence. If a sub-population no longer produces new winners during the convergence process, the sub-population is said to be mature. When the sub-population matures, the convergence process of the sub-population ends;

[0076] ③ Alienation operation of subpopulations: After the convergence operation of each winning subpopulation and temporary subpopulation is completed, the alienation operation is performed. In the entire solution space, each subpopulation competes to become the winner and constantly detects the point in the center of the solution space. This process is called alienation;

[0077] ④ Determine whether the termination condition is met: When the conditions for stopping the iteration are met, the algorithm ends the optimization process. If not, it returns to step ②;

[0078] ⑤ Output the optimal solution: According to the coding rules, analyze the optimal individual found and output the optimal solution.

[0079] In this embodiment, in view of the shortcomings of the BP neural network algorithm, the MEA algorithm is used to optimize the weights and values ​​of the BP neural network, and a MEA-BP prediction model is established. The main steps of the MEA algorithm to optimize the BP neural network are as follows:

[0080] (1) According to the BP neural network topology, the solution space is mapped to the encoding space, and the encoding length s is: s = nl + ml + l + m (2), where n is the number of input nodes; m is the number of output nodes; l is the number of hidden layer nodes;

[0081] (2) Set the maximum number of iterations i max ;

[0082] (3) Select the score function v for each individual and population, Where A2 is the output value of the output layer after each iteration; T is the expected output; N SE is the mean square error; R SE is the root mean square; A2 i is the output value of the output layer after the i-th iteration; N is the number of iterations; since the population needs to screen the value with the highest score, the fitness function v takes the inverse of the mean square error between the expected value and the actual output value of each iteration;

[0083] (4) Continuously iterate and output the optimal individual, which is used as the network weight and threshold to train the network. The MEA-BP neural network algorithm process is as follows: Figure 8 shown.

[0084] In this embodiment, the present invention can also adopt a battery energy storage system state of charge (SOC) prediction method based on MEA-BP neural network to accurately evaluate the SOC state of the energy storage battery, select the battery external characteristic current, voltage and temperature that are easier to obtain as the input of the BP neural network, and the SOC state as the output of the neural network, and use the mind evolution algorithm (MEA) to optimize the weights and thresholds of the BP network to evaluate the SOC state of the battery energy storage system. The algorithm of the present invention can not only improve the prediction accuracy, but also accelerate the learning speed of the BP neural network, and can realize accurate estimation of the SOC of the battery energy storage system.

[0085] The present invention is a battery energy storage system state evaluation system and method. The main factors affecting the SOC operating state of the battery energy storage system are analyzed, and a battery state of charge SOC evaluation method based on a GA-BP neural network model based on data mining is adopted. In use, based on the energy storage system charge and discharge test, the present invention analyzes a large number of battery energy storage system operation history data, uses rough set theory to identify various factor variables affecting SOC prediction, simplifies attributes to determine the input variables of the neural network prediction model, and proposes a rough set-neural network (GA-BP) battery SOC estimation method based on data mining technology. First, the rough set theory is used to simplify the attributes of the main factors affecting SOC, and the key characteristic parameters that can correctly evaluate the battery SOC are extracted; then the above parameters are used as the input of the GA-BP neural network to establish a battery energy storage system SOC prediction model based on GA-BP, so as to further improve the prediction accuracy of the energy storage system SOC; the present invention has the advantages of data mining based on rough set theory, using neural network to realize state evaluation, and improving prediction accuracy.

Claims

1. A battery energy storage system status assessment system, comprising an energy storage battery management system, characterized in that: The energy storage battery management system includes an energy storage battery system, a MATLAB simulation module, a core assembly module, a MySQL database, and a WEB management demonstration platform; the MATLAB simulation module uses a battery model to simulate a real energy storage battery system, can output battery status parameters, and can accept input and change status parameters according to the input; the core assembly module is responsible for the core calculation of the battery management system, including; the MySQL database is used to store real-time updated battery parameters, user information, and status parameters and auxiliary data of the core assembly module; the Web management demonstration platform is used to provide an entrance to the battery management system, so as to facilitate real-time understanding of the system's operating status and issue corresponding instructions.

2. A battery energy storage system state assessment method, which adopts a battery energy storage system SOC state assessment based on data mining technology and is applied to a battery energy storage system state assessment system as claimed in claim 1, characterized in that: The method comprises the following steps: Step 1: Battery energy storage system charging and discharging test process; Step 2: Attribute reduction method based on rough set theory; Step 3: GA-BP neural network battery SOC state evaluation method based on rough set attribute reduction.

3. A battery energy storage system status assessment method as claimed in claim 2, characterized in that: The battery energy storage system charge and discharge test process in step 1 includes the following steps: Step 1.1: Battery SOC: SOC represents the remaining battery power or remaining operating time in the current charging cycle. The ratio of the remaining capacity during battery operation to the rated capacity of the battery is usually called the battery SOC: In the formula, Q C is the remaining capacity of the battery; Q I is the rated capacity of the battery; Step 1.2: Battery variable current intermittent charging and discharging process: The collected data include the battery current I, voltage U, internal resistance R, power P, cycle power change W1, cycle charge W2, cycle discharge W3, battery temperature T and battery remaining capacity Q at each moment. The SOC value of the entire moment is calculated using the remaining capacity value Q at each moment according to formula (1).

4. A battery energy storage system status assessment method according to claim 2, characterized in that: The attribute simplification method based on rough set theory in step 2 is specifically as follows: starting from the empty set, calculating the attribute importance of the remaining attributes each time, selecting attributes with large attribute importance and incorporating them into the simplified set until the simplification conditions are met, ensuring that important attributes are added to the simplification first, so as to facilitate rapid finding of the simplification.

5. A battery energy storage system status assessment method according to claim 4, characterized in that: The reduced set in step 2 is: according to the importance of the conditional attribute, the reduced set can be obtained by judging whether the attribute importance is greater than zero in the neighborhood rough set reduction.

6. A battery energy storage system status assessment method according to claim 2, characterized in that: The GA-BP neural network battery SOC state evaluation method based on rough set attribute reduction in step 3 comprises the following steps: Step 3.1: Construct a SOC prediction model based on BP neural network; Step 3.2: GA-BP neural network; Step 3.3: Battery SOC state evaluation method based on RS-GA-BP neural network.

7. A method for evaluating the state of a battery energy storage system according to claim 6, characterized in that: The construction of the SOC prediction model based on the BP neural network in step 3.1 is specifically as follows: the working voltage U, current I and temperature T of the battery energy storage system at a certain moment are the main factors affecting the SOC, so these three factors are used as the input vector [X1, X2, X3] of the model, that is, the number of input nodes of the neural network is 3, and the battery SOC is used as the output vector [Y] of the model, that is, the number of output nodes is 1; 12 nodes are used in the hidden layer to accurately describe the relationship between the battery voltage, current, temperature and the battery SOC.

8. A method for evaluating the state of a battery energy storage system according to claim 6, characterized in that: The GA-BP neural network in step 3.2 includes the following steps: Step 3.21: Generate the initial population step; Step 3.22: Determine the relevant parameters and termination conditions of the genetic algorithm, initialize the genetic algorithm and perform calculations; Step 3.23: Evaluate the calculation results, decode the BP network and calculate the fitness value of each chromosome; Step 3.24: Determine whether the genetic algorithm has reached the termination condition. If so, the calculation ends and the algorithm jumps to step 3.26 for execution. If not, the algorithm continues to execute downward. Step 3.25: Perform selection, crossover, and mutation genetic algorithm operations to generate a new generation of chromosomes, and then go to step 3.23 for a loop operation; Step 3.26: Use the BP network to locally fine-tune the approximate optimal value searched by the genetic algorithm to improve the accuracy of the solution until the accuracy requirements are met and proceed to the next step; Step 3.27: Save the BP network model calculated by the sample and input the test sample for comparative analysis; Step 3.28: Output the results.

9. A method for evaluating the state of a battery energy storage system according to claim 6, characterized in that: The battery SOC state evaluation method based on RS-GA-BP neural network in step 3.3 comprises the following steps: Step 3.31: Use rough sets as a precondition system; Step 3.32: Preprocess the samples before SOC state prediction by removing redundant attributes and identical samples through attribute simplification; Step 3.33: Then send the data after neighborhood rough set preprocessing into the GA-BP neural network for prediction.

Citation Information

Cited By

  • Storage battery intelligent monitoring system

    CN120009744A

  • An intelligent monitoring system for storage batteries

    CN120009744B