Improved ampere-hour integral SOC estimation method based on reconfigurable battery topology
By constructing a BP neural network based on genetic algorithm optimization, establishing the mapping relationship between SOC and OCV and making corrections, the problem of insufficient accuracy of traditional SOC estimation methods under the uncertainty of coulombic efficiency and battery capacity is solved, the accuracy of the battery management system is improved and the hardware requirements are simplified.
Patent Information
- Application Number
- CN202510925005.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-04
- Publication Date
- 2025-10-17
AI Technical Summary
Traditional SOC estimation methods are not accurate enough due to the uncertainty of coulombic efficiency and battery capacity, and do not fully utilize the high flexibility of reconfigurable battery topology.
A BP neural network optimized based on genetic algorithm is constructed to establish the SOC and OCV mapping relationship after the battery is disconnected for 1s in the topology structure. The SOC estimation value is corrected by combining the GA-BP neural network, and a correction factor is introduced to offset the influence of coulombic efficiency and battery capacity uncertainty.
The accuracy of SOC estimation is improved, hardware requirements are reduced, system complexity is simplified, and real-time correction of battery SOC is achieved, thereby enhancing the overall performance of the battery management system.
Smart Images
Figure CN120802043A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application belongs to the technical field of battery energy storage, in particular to the battery state estimation technology, and relates to an SOC (State of Charge) estimation method based on a reconfigurable topology. BACKGROUND
[0002] Battery state estimation is an important technical link in the battery management system, and accurate estimation of SOC is of great significance to prolonging the battery life and improving the system efficiency. Traditional SOC estimation methods, such as ampere-hour integration, are easily affected by the uncertainty of coulomb efficiency and battery capacity during the estimation process, resulting in deviation of the estimation results. Moreover, most of the existing SOC estimation methods do not fully utilize the high flexibility characteristics of the reconfigurable battery topology. In order to improve the accuracy of battery SOC estimation in the reconfigurable battery topology, an improved ampere-hour integration SOC estimation method based on the reconfigurable battery topology is proposed. SUMMARY
[0003] The application provides an improved ampere-hour integration SOC estimation method based on a reconfigurable topology, which can utilize the characteristics of the reconfigurable battery topology to establish the SOC-OCV mapping relationship between the open-circuit voltage value after the battery is disconnected from the circuit for 1 second and the discharge rate. The method corrects the SOC estimation value by combining the GA-BP neural network, thereby improving the accuracy of the estimation.
[0004] To achieve the above-mentioned purpose, the application provides an improved ampere-hour integration SOC estimation method based on a reconfigurable battery topology, a BP neural network optimized based on a genetic algorithm (GA-BP neural network) is constructed to fit the mapping relationship between the SOC and the open-circuit voltage (OCV) of the battery after being disconnected from the topology structure for 1 second, and the SOC value estimated by the ampere-hour integration method is corrected based on the mapping relationship. The specific steps are as follows:
[0005] S1. Collect the SOC and OCV data of the battery after being disconnected for 1 second under multiple discharge rates, temperatures and cycle times, establish a GA-BP neural network model and construct a SOC-OCV mapping database;
[0006] S2. Estimate the SOC state of the battery in real time by using the ampere-hour integration method when the battery is discharging;
[0007] S3. If the SOC decreases by more than ΔSOC and it is judged that the current battery meets the condition of being disconnected from the reconfigurable topology, call the SOC-OCV curve matching the current condition through the database and perform the correction operation; otherwise, delay the correction, wherein ΔSOC is the SOC change amount;
[0008] S4. Determine whether the discharge or charging is finished, if yes, stop the estimation; if not, execute S2-S4 in turn.
[0009] For the establishment of GA-BP neural network model and the construction of SOC-OCV mapping database in S1, the following steps are included:
[0010] (1) Control each battery to discharge at an interval of 5% discharge capacity, and record the open circuit voltage value at this time, the discharge rate before disconnection, temperature and cycle number after each discharge for 1s;
[0011] (2) The collected data are used as the training set of GA-BP neural network, and the short-time static SOC-OCV mapping relationship under the influence of discharge rate, temperature and cycle number is established;
[0012] (3) The established model is used to predict the SOC-OCV relationship under the remaining discharge rate, temperature and cycle number.
[0013] The correction factor μ is introduced k to offset the influence of coulomb efficiency and battery capacity uncertainty on the accuracy of SOC estimation by ampere-hour integral method, and the improved ampere-hour integral method is shown in formula (1).
[0014]
[0015] wherein, SOC Ah (t-1) is the estimated value at the last time, SOC Ah (t) is the estimated value at the current time, η is the coulomb efficiency, C e is the battery capacity, I is the battery discharge current at t time, μ k is the correction factor, and the initial value is 1.
[0016] Under the premise that the short-time accuracy meets the requirements, only when the estimated SOC value by the improved ampere-hour integral method decreases by 5%, formula (1) is modified by formula (2).
[0017] SOC Ah (t) = SOC OCV (t) (2)
[0018] wherein, SOC OCV (t) is the SOC-OCV mapping curve at 1s after disconnection fitted by GA-BP neural network.
[0019] The difference between SOC Ah (t) and SOC OCV (t) is used to determine the deviation trend of the SOC estimation error by ampere-hour integral method, and the correction factor in formula (1) is modified by formula (3).
[0020] μ k = μ k-1 - α * (SOC OCV (t) - SOC Ah (t)) (3)
[0021] Wherein, alpha is proportional gain constant.
[0022] The technical effect of the present application is:
[0023] Compared with the traditional SOC estimation method, the present application introduces GA-BP neural network and SOC-OCV mapping relationship, so that the SOC estimation is more accurate. By using the reconfigurable battery topology, the battery is disconnected from the circuit for a short time, which can effectively reduce the hardware demand and simplify the system complexity. In addition, the improved ampere-hour integral method realizes real-time correction of battery SOC estimation on the basis of low error, and enhances the overall performance of the battery management system. BRIEF DESCRIPTION OF DRAWINGS
[0024] The accompanying drawings, which form a part of this application, are included to provide a further understanding of the application and are incorporated in and constitute a part of this application. The embodiments of this application and of the related description are used to explain the application and are not intended to limit the application. In the drawings:
[0025] Figure 1 It is a reconfigurable battery topology in the embodiment of the present application;
[0026] Figure 2 It is an improved ampere-hour integral method SOC estimation flowchart in the embodiment of the present application;
[0027] Figure 3 It is an OCV and SOC curve graph after the battery is disconnected for 1 second in the embodiment of the present application;
[0028] Figure 4 It is a GA-BP neural network SOC prediction flowchart in the embodiment of the present application;
[0029] Figure 5 It is a GA-BP prediction 0.5C discharge rate SOC-OCV error graph in the embodiment of the present application;
[0030] Figure 6 It is a comparison graph of each SOC estimation method in the embodiment of the present application;
[0031] Figure 7 It is an error comparison graph of each SOC estimation method in the embodiment of the present application. DETAILED DESCRIPTION
[0032] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other in the case of no conflict. The present application will be described in detail below with reference to the drawings and in combination with the embodiments.
[0033] It should be noted that the steps shown in the flowchart of the drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described herein can be executed in an order different from that herein.
[0034] In view of the deficiencies in the prior art, an improved ampere-hour integral SOC estimation method based on a reconfigurable topology is provided herein, which can utilize the characteristics of the reconfigurable battery topology to establish a SOC-OCV mapping relationship between the open circuit voltage value after the battery is disconnected from the circuit for 1 second and the discharge rate. The method corrects the SOC estimation value by combining GA-BP neural network, thereby improving the accuracy of the estimation.
[0035] The present application is realized by the following technical solutions:
[0036] The present application provides an improved ampere-hour integral SOC estimation method based on a reconfigurable battery topology, constructs a BP neural network (GA-BP neural network) based on a genetic algorithm optimization, which is used to fit the mapping relationship between the SOC and the open circuit voltage (OCV) of the battery after being disconnected from the topology structure for 1s, and corrects the SOC value estimated by the ampere-hour integral method based on the mapping relationship. The specific steps are as follows:
[0037] S1. Collect the SOC and OCV data of the battery after being disconnected for 1s under multiple discharge rates, temperatures and cycle times, establish a GA-BP neural network model and construct a SOC-OCV mapping database;
[0038] S2. Estimate the SOC state of the battery in real time by using the ampere-hour integral method when the battery is discharging;
[0039] S3. If the SOC decreases by more than ΔSOC, and it is judged that the current battery meets the condition of being disconnected from the reconfigurable topology, then the SOC-OCV curve matching the current condition is called through the database, and the correction operation is performed; otherwise, the correction is delayed, wherein ΔSOC is the SOC change;
[0040] S4. Judge whether the discharging or charging is finished, if yes, stop the estimation; if not, execute S2-S4 in turn.
[0041] For S1, the GA-BP neural network model is established and the SOC-OCV mapping database is constructed, including the following steps:
[0042] (1) Control each battery to discharge at intervals of 5% of the discharge capacity, and stand for 1s after each discharge, and record the open-circuit voltage value at that time and the discharge rate, temperature and cycle number before disconnection;
[0043] (2) The collected data is used as the training set of the GA-BP neural network, and the short-time standing SOC-OCV mapping relationship considering the influence of discharge rate, temperature and cycle number is established;
[0044] (3) The established model is used to predict the SOC-OCV relationship under the remaining discharge rate, temperature and cycle number.
[0045] A correction factor μ is introduced k To offset the influence of coulomb efficiency and battery capacity uncertainty on the accuracy of SOC estimation by ampere-hour integration method, the improved ampere-hour integration method is shown in formula (4).
[0046]
[0047] Wherein, SOC Ah (t-1) is the estimated value at the last time, SOC Ah (t) is the estimated value at the current time, η is the coulomb efficiency, C e is the battery capacity, I is the battery discharge current at t, μ k is the correction factor, and the initial value is 1.
[0048] Under the premise that the short-time accuracy meets the requirements, only when the estimated SOC value by the improved ampere-hour integration method decreases by 5%, formula (4) is modified by formula (5).
[0049] SOC Ah (t) = SOC OCV (t) (5)
[0050] Wherein, SOC OCV (t) is the SOC-OCV mapping curve at 1s after disconnection fitted by GA-BP neural network.
[0051] At the same time, the difference between SOC Ah (t) and SOC OCV (t) is used to judge the deviation trend of the SOC estimation error by ampere-hour integration method, and the correction factor in formula (4) is modified by formula (6).
[0052] μ k = μ k-1 - α * (SOC OCV (t) - SOC Ah (t)) (6)
[0053] Wherein, α is the proportional gain constant.
[0054] Example
[0055] like Figures 1-7 As shown in FIG, the present invention provides an improved ampere-hour integral SOC estimation method based on a reconfigurable battery topology, constructs a BP neural network (GA-BP neural network) based on genetic algorithm optimization, and is used to fit the mapping relationship between the SOC and the open circuit voltage (OCV) of the battery after being disconnected for 1 second in the topology structure, and corrects the SOC value estimated by the ampere-hour integral method based on the mapping relationship, as shown in FIG. Figure 2 The specific steps are as follows:
[0056] S1. Collect SOC and OCV data after battery disconnection for 1 second at various discharge rates, temperatures, and cycle times, establish a GA-BP neural network model, and construct a SOC-OCV mapping database;
[0057] S2. Use the ampere-hour integration method to estimate the battery SOC in real time during battery discharge.
[0058] S3. If the SOC drops by more than 5% and the battery is deemed to be disconnected from the reconfigurable topology, the SOC-OCV curve matching the current conditions is retrieved from the database and a correction is performed; otherwise, the correction is delayed.
[0059] S4. Determine whether the discharge or charge is completed. If so, stop estimating. If so, execute S2-S4 in sequence.
[0060] The establishment of the GA-BP neural network model and the construction of the SOC-OCV mapping database in S1 include the following steps:
[0061] (1) Figure 3 As shown, each battery is controlled to discharge at intervals of 5% of the discharge capacity, with discharge rates of 0.2C, 0.4C, 0.6C, 0.8C and 1C, and allowed to stand for 1s after each discharge. The open circuit voltage at that moment and the discharge rate, temperature and number of cycles before disconnection are recorded.
[0062] (2) Figure 4 As shown in the figure, the collected data is used as the training set of the GA-BP neural network to establish a short-time static SOC-OCV mapping relationship considering the influence of discharge rate, temperature and cycle number;
[0063] (3) The established model is used to predict the SOC-OCV relationship under 0.5C discharge rate, temperature and cycle number. The error is shown in the figure below. Figure 5 shown.
[0064] Introducing the correction factor μ kTo offset the influence of coulomb efficiency and battery capacity uncertainty on the accuracy of SOC estimated by ampere-hour integral method, the ampere-hour integral method is improved as shown in formula (7).
[0065]
[0066] Wherein, SOC Ah (t-1) is an estimated value at the last time, SOC Ah (t) is an estimated value at the current time, η is coulomb efficiency, C e is battery capacity, I is battery discharge current at t, μ k is a correction factor, and the initial value is 1.
[0067] Under the premise that short-time accuracy meets the requirements, only when the estimated SOC value of the improved ampere-hour integral method decreases by 5%, formula (7) is corrected by formula (8).
[0068] SOC Ah (t) = SOC OCV (t) (8)
[0069] Wherein, SOC OCV (t) is an SOC-OCV mapping curve at 1s obtained by fitting GA-BP neural network.
[0070] At the same time, the difference between SOC Ah (t) and SOC OCV (t) is used to determine the deviation trend of the SOC estimation error of the ampere-hour integral method, and the correction factor in formula (7) is corrected by formula (9).
[0071] μ k = μ k-1 - α * (SOC OCV (t) - SOC Ah (t)) (9)
[0072] Wherein, α is a proportional gain constant. When SOC Ah (t) is greater than SOC OCV (t), it indicates that the influence of coulomb efficiency and battery capacity causes the value obtained by the ampere-hour integral method to be larger than the actual value, and μ k is increased by formula (9) to compensate for the error caused by coulomb efficiency and battery capacity.
[0073] Based on MATLAB / Simulink, the SOC estimation value of a single battery is simulated, and it is shown in Figure 6 and Figure 7 that the improved ampere-hour integral SOC estimation method based on the reconfigurable battery topology proposed in the application can improve the accuracy of battery SOC estimation when the battery works in the reconfigurable battery topology.
[0074] The above description is only preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.
Claims
1. An improved ampere-hour integral SOC estimation method based on reconfigurable battery topology, characterized in that: A BP neural network optimized by genetic algorithm (GA-BP neural network) is constructed to fit the mapping relationship between the SOC and open circuit voltage (OCV) of the battery after 1 second of disconnection in the topology structure, and based on this mapping relationship, the SOC value estimated by the ampere-hour integration method is corrected. The specific steps are as follows: S1. Collect SOC and OCV data after battery disconnection for 1 second at various discharge rates, temperatures, and cycle times, establish a GA-BP neural network model, and construct a SOC-OCV mapping database; S2. Use the ampere-hour integration method to estimate the battery SOC in real time during battery discharge. S3. If the SOC drops by more than ΔSOC and it is determined that the current battery meets the conditions for disconnection from the reconfigurable topology, the SOC-OCV curve that matches the current conditions is called from the database and a correction operation is performed; Otherwise, delay correction, where ΔSOC is the SOC change; S4. Determine whether the discharge or charge is completed. If so, stop estimating. If so, execute S2-S4 in sequence.
2. S1 according to claim 1, characterized in that The following steps are involved: (1) Each battery is discharged at intervals of 5% of its discharge capacity, and the battery is left to stand for 1 second after each discharge. The open circuit voltage at that moment and the discharge rate, temperature, and number of cycles before disconnection are recorded. (2) The collected data are used as the training set of the GA-BP neural network to establish a short-time static SOC-OCV mapping relationship considering the influence of discharge rate, temperature and cycle number; (3) Use the established model to predict the SOC-OCV relationship under other discharge rates, temperatures and cycle numbers.
3. S3 according to claim 1, characterized in that Introducing the correction factor μ k To offset the influence of coulomb efficiency and battery capacity uncertainty on the accuracy of SOC estimation using the ampere-hour integration method, the improved ampere-hour integration method is shown in formula (1). Among them, SOC Ah (t-1) is the estimated value at the previous moment, SOC Ah (t) is the estimated value at the current moment, η is the Coulomb efficiency, C e is the battery capacity, I is the battery discharge current at time t, μ k is the correction factor, and its initial value is 1. On the premise that the short-time accuracy meets the requirements, formula (1) is corrected by formula (2) only when the SOC value estimated by the improved ampere-hour integration method drops by 5%. SOC Ah (t)=SOC OCV (t) (2) Among them, SOC OCV (t) is the SOC-OCV mapping curve fitted by the GA-BP neural network when disconnected for 1s. While using SOC Ah (t) and SOC OCV The difference between (t) and (t) is used to judge the error deviation trend of the SOC estimation using the ampere-hour integration method, and the correction factor in formula (1) is corrected using formula (3). μ k =μ k-1 -α*(SOC OCV (t)-SOC Ah (t)) (3) Where α is the proportional gain constant.