Battery SOC correction calibration method and system

Abstract

The invention provides a battery SOC correction calibration method and system, and the method comprises the steps: obtaining a characteristic value representing the characteristics of a battery according to the charging data through independently charging a single cluster of batteries in an industrial and commercial energy storage cabin, building a battery model comprising a BP neural network model, and carrying out the correction calibration of the SOC of the battery according to the current charging condition and the characteristic value of a background battery. And the estimated SOC is obtained through the SOC correction module. According to the invention, the technical problems of low SOC measurement accuracy, poor real-time performance and low charging and high discharging are solved.

Description

Technical Field

[0001] The present invention relates to battery management technology, and in particular to a battery SOC correction and calibration method and system. Background Art

[0002] Industrial and commercial energy storage systems are widely used in power peak regulation, new energy grid connection, electric vehicle charging stations and other fields. Accurate calibration of battery SOC is the key to ensure the safe and stable operation of energy storage systems. However, due to the complexity of the chemical reactions inside the battery and the diversity of the working environment, the battery SOC is difficult to measure accurately. At the same time, the actual SOC of multiple clusters of batteries is different, and there is a problem of low charging and high discharge. Traditional SOC calibration methods, such as open circuit voltage method and ampere-hour measurement method, have problems such as low accuracy and poor real-time performance, and cannot solve the problem of high charging and low discharge.

[0003] The existing invention patent application document with publication number CN114089203A, "A method for automatic calibration and SOC estimation of an electrochemical energy storage system", includes: realizing automatic calibration of battery capacity and acquisition of battery operating characteristic curves through linkage control among BMS, PCS, and EMS. After BMS acquires the calibrated battery capacity and battery operating characteristic curve, it automatically updates the battery capacity value and battery operating characteristic curve in the database. During the operation of the electrochemical energy storage system, BMS estimates SOC based on the latest acquired battery capacity and battery operating characteristic curve. However, the aforementioned solution cannot solve the cluster balance problem of battery replenishment. After correcting the SOC, the charging of the entire cabin still stops at the highest cluster SOC, and the same is true for discharge.

[0004] The existing invention patent application document "An Energy Storage System SOC Online Correction System" with publication number CN107957558A, the existing system includes: an energy storage system, a DC voltage detector, a voltage mutation judgment module, a correction calculation module, and an SOC online correction module; the energy storage system includes multiple batteries, the DC voltage detector is arranged at the output end of the energy storage system, and is used to obtain the DC side voltage of the energy storage system battery; the voltage mutation judgment module is provided with a voltage change rate threshold, when the DC side voltage exceeds the voltage change rate threshold, the SOC online correction module is started; the correction calculation module is used to obtain the correction amount of the energy storage system SOC according to the DC side voltage change rate and the power output value of the energy storage system; the SOC online correction module corrects the SOC of the energy storage system based on the calculated correction amount. The published scheme judges the battery SOC by calculating the voltage mutation, and performs battery SOC cycle balance by connecting the batteries in parallel. It is impossible to predict the correction value of the single cluster SOC through the changes in voltage, current, and temperature during the charging process recorded by the BMS.

[0005] In summary, the prior art has technical problems such as low SOC measurement accuracy, poor real-time performance, and low charging and high discharging. Summary of the invention

[0006] The technical problem to be solved by the present invention is: how to solve the technical problems of low SOC measurement accuracy, poor real-time performance and low charging and high discharging in the prior art.

[0007] The present invention adopts the following technical solution to solve the above technical problem: A battery SOC calibration method comprises:

[0008] S1. Maintain the normal operation of the battery pack, battery management system BMS and related equipment; obtain and control the test environment conditions based on the test environment information;

[0009] S2, performing charging operation and data collection operation, performing constant current-constant voltage charging on each battery cluster in the battery pack separately, and recording battery parameters in real time during the charging operation to form charging data;

[0010] S3, performing a feature value extraction operation on the charging data to extract a feature value of the battery charging characteristic;

[0011] S4. Building a battery model based on the characteristic value of the battery charging characteristic, wherein the battery model includes: a BP neural network model, enabling the BP neural network model to learn according to the historical charging data to train and obtain a suitable prediction model, using the suitable prediction model to simulate the charging and discharging process of the battery, and predicting the SOC value prediction result of the battery;

[0012] S5, perform SOC online correction operation, obtain and correct the SOC value prediction result through the SOC correction module according to the current charging status of the background battery and the characteristic value of the battery charging characteristics, and obtain the corrected SOC estimated value;

[0013] S6, performing a charging operation according to the corrected SOC estimated value until the battery of the current cluster reaches a fully charged state;

[0014] S7, performing the calibration operation of S5 and the charging operation of S6 on the batteries of all the clusters in the battery pack until the calibration operation and the charging operation of the battery pack are completed.

[0015] The present invention performs an online SOC correction and calibration operation on a single-cluster battery of an industrial and commercial energy storage system. Each cluster of batteries is charged individually, and characteristic values ​​characterizing battery characteristics are obtained according to the charging data of the single-cluster battery. A battery model including a BP neural network model is built, and then an estimated SOC is obtained through an SOC correction module according to the current charging status and characteristic values ​​of the background battery, thereby realizing online correction of the SOC of the single-cluster battery, and replenishing the SOC of each cluster of batteries according to the correction result.

[0016] The present invention can obtain the SOC of the battery in real time through online correction, thereby improving the operating efficiency of the energy storage system. The present invention can be applied to industrial and commercial energy storage systems of different scales and configurations, and has strong versatility and scalability.

[0017] In a more specific technical solution, in S2, the battery parameters include: voltage V(t), current I(t) and temperature T(t).

[0018] In a more specific technical solution, in S3, the battery charging characteristic feature values ​​include: charging capacity Qc, charging efficiency η and voltage curve slope k.

[0019] In a more specific technical solution, in S3, the method for obtaining the characteristic value of the battery charging characteristic includes: polynomial fitting, spline function fitting, charging parameter analysis and voltage curve analysis.

[0020] In more specific technical solutions, S3 includes:

[0021] S31. Calculate the charging capacity Qc received by the battery during the charging operation using the following logic:

[0022] Qc=∫I(t)dt;

[0023] S32. Calculate the charging efficiency η of the battery during the charging operation using the following logic:

[0024] η = (final power - initial power) / (charged power));

[0025] S33. Calculate the voltage curve slope k using the following logic:

[0026] k = dV(t) / dt.

[0027] In a more specific technical solution, S4 includes:

[0028] S41, constructing a BP neural network model, wherein the BP neural network model includes: an input layer, a hidden layer and an output layer;

[0029] S42, determining the number of hidden layers and the number of neurons, and optimizing the network structure of the BP neural network model through preset test operations;

[0030] S43, selecting an applicable activation function and an applicable loss function;

[0031] S44. Use the historical charging data and its corresponding SOC value to train the BP neural network model, adjust the network weights and thresholds through the back propagation algorithm, and obtain a suitable network model.

[0032] The present invention records the changes in voltage, current and temperature during the charging process through the BMS and predicts and calculates the correction value of the single cluster SOC through a learning algorithm. The present invention selects appropriate activation functions and loss functions, such as Sigmoid functions and mean square error functions, to improve the prediction accuracy of the model.

[0033] The present invention utilizes a large amount of historical charging data and corresponding SOC values ​​to train the BP neural network model, and adjusts the network weights and thresholds through the back propagation algorithm so that the model can accurately simulate the charging and discharging process of the battery.

[0034] In a more specific technical solution, in S41, the input layer is used to receive the characteristic value of the battery charging characteristic as input data, the input data is processed to obtain a predicted SOC value, and the output layer is used to output the SOC value prediction result.

[0035] In a more specific technical solution, in S5, the SOC correction module uses the following logic to correct the SOC value prediction result:

[0036] SOC_corrected=SOC_predicted*f(Qc,η,k,...)

[0037] Where f is the function that is corrected according to the eigenvalue.

[0038] The present invention utilizes the BP neural network model to simulate the battery charging and discharging process, combines the characteristic value to perform SOC correction, and improves the estimation accuracy of SOC.

[0039] The present invention corrects the prediction result of the BP neural network model through the SOC correction module according to the current charging status and characteristic value of the background battery, so as to obtain a more accurate SOC estimation value.

[0040] In a more specific technical solution, S7 includes:

[0041] S71, performing inter-cluster switching and overall calibration operations on the battery group, and disconnecting the current cluster when a cluster of batteries is fully charged;

[0042] S72, executing S71 for the next battery cluster, connecting and calibrating, until the SOC calibration and full charging operations of the batteries in all clusters are completed.

[0043] In a more specific technical solution, a battery SOC calibration system includes:

[0044] Equipment and experimental condition setting module, used to maintain the normal operation of battery pack, battery management system BMS and related equipment; obtain and control the test environment conditions according to the test environment information;

[0045] The charging and data acquisition module is used to perform charging and data acquisition operations. Each battery cluster in the battery pack is charged with constant current and constant voltage individually. During the charging operation, the battery parameters are recorded in real time to form charging data. The charging and data acquisition module is connected to the equipment and experimental condition setting module.

[0046] A feature value extraction module is used to perform feature value extraction operation on charging data to extract the feature value of battery charging characteristics. The feature value extraction module is connected to the charging and data acquisition module;

[0047] A model building and training prediction module is used to build a battery model based on the characteristic value of the battery charging characteristic, wherein the battery model includes: a BP neural network model, which enables the BP neural network model to learn according to historical charging data to train a suitable prediction model, and uses the suitable prediction model to simulate the charging and discharging process of the battery to predict the SOC value prediction result of the battery. The model building and training prediction module is connected to the characteristic value extraction module;

[0048] The SOC correction module is used to perform SOC online correction operations, obtain and correct the SOC value prediction result according to the current charging status of the background battery and the characteristic value of the battery charging characteristics through the SOC correction module to obtain the corrected SOC estimation value. The SOC correction module is connected to the characteristic value extraction module;

[0049] A charging module, used for performing a charging operation according to the corrected SOC estimation value until the battery of the current cluster reaches a fully charged state, and the charging module is connected to the SOC correction module;

[0050] The battery pack cycle correction and charging module is used to perform the correction operation of S5 and the charging operation of S6 on the batteries of all clusters in the battery pack until the battery pack completes the correction operation and the charging operation. The battery pack cycle correction and charging module is connected to the charging module and the SOC correction module.

[0051] Compared with the prior art, the present invention has the following advantages:

[0052] The present invention performs online calibration and correction of the SOC of a single cluster battery in an industrial and commercial energy storage system. Each cluster battery is charged separately, and characteristic values ​​characterizing the battery characteristics are obtained according to the charging data of the single cluster battery. A battery model including a BP neural network model is built, and then an estimated SOC is obtained through an SOC correction module according to the current charging status and characteristic values ​​of the background battery, thereby realizing online calibration of the SOC of a single cluster battery, and replenishing the SOC of each cluster battery according to the calibration results. The present invention records the charging process of the BMS, replenishes the single cluster battery with constant current and constant voltage, records the BMS data during the replenishment process, corrects the SOC value through an algorithm, and then performs the remaining SOC replenishment operation.

[0053] The present invention can obtain the SOC of the battery in real time through online correction, thereby improving the operating efficiency of the energy storage system. The present invention can be applied to industrial and commercial energy storage systems of different scales and configurations, and has strong versatility and scalability.

[0054] The present invention selects appropriate activation function and loss function, such as Sigmoid function and mean square error function, to improve the prediction accuracy of the model.

[0055] The present invention utilizes a large amount of historical charging data and corresponding SOC values ​​to train the BP neural network model, and adjusts the network weights and thresholds through the back propagation algorithm so that the model can accurately simulate the charging and discharging process of the battery.

[0056] The present invention utilizes the BP neural network model to simulate the battery charging and discharging process, combines the characteristic value to perform SOC correction, and improves the estimation accuracy of SOC.

[0057] The present invention corrects the prediction result of the BP neural network model through the SOC correction module according to the current charging status and characteristic value of the background battery, so as to obtain a more accurate SOC estimation value.

[0058] The present invention solves the technical problems of low SOC measurement accuracy, poor real-time performance and low charging and high discharging in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Schematic diagram of the basic steps of a battery SOC calibration method according to Embodiment 1 of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] Example 1

[0062] like Figure 1 As shown, a battery SOC calibration method provided by the present invention includes the following basic steps:

[0063] S1. During the preparation, conduct a comprehensive inspection of the industrial and commercial energy storage system to ensure the normal operation of the battery pack, battery management system (BMS) and other related equipment; at the same time, ensure that the temperature, humidity and other conditions of the test environment meet the requirements;

[0064] S2. During the charging and data collection process, each battery cluster is charged with constant current and constant voltage individually. During the charging process, the battery parameters are recorded in real time to form charging data;

[0065] In this embodiment, the battery parameters include but are not limited to: voltage V(t), current I(t) and temperature T(t);

[0066] S3, in the feature value extraction process, extracting feature values ​​representing battery characteristics according to charging data;

[0067] In this embodiment, the method for obtaining the characteristic value characterizing the battery characteristics includes but is not limited to: polynomial fitting, spline function fitting and signal analysis, which are used to describe the charging characteristics of the battery;

[0068] In this embodiment, the charging capacity Qc represents the total amount of electricity received by the battery during the charging process:

[0069] Qc=∫I(t)dt;

[0070] The charging efficiency η represents the energy conversion efficiency of the battery during the charging process:

[0071] η = (final power - initial power) / (charged power));

[0072] The slope k of the voltage curve indicates how fast the voltage changes over time:

[0073] k = dV(t) / dt;

[0074] S4. In the process of building the battery model, a battery model is built based on the extracted characteristic values. The battery model includes a BP neural network model. The BP neural network model can simulate the charging and discharging process of the battery and predict the SOC of the battery by learning a large amount of data;

[0075] In this embodiment, a BP neural network model is constructed, which includes but is not limited to: an input layer, a hidden layer, and an output layer. Specifically, the input layer receives a feature value as input, and the output layer outputs a predicted SOC value. The number of layers and neurons in the hidden layer is determined, and the network structure is optimized through experiments and adjustments.

[0076] In this embodiment, appropriate activation function and loss function are selected, such as Sigmoid function and mean square error function, to improve the prediction accuracy of the model.

[0077] In this embodiment, a large amount of historical charging data and corresponding SOC values ​​are used to train the BP neural network model, and the network weights and thresholds are adjusted through the back propagation algorithm so that the model can accurately simulate the charging and discharging process of the battery;

[0078] S5, perform SOC online correction, according to the current charging status and characteristic value of the background battery, correct the prediction result of the BP neural network model through the SOC correction module to obtain a more accurate SOC estimation value;

[0079] S6, replenishing the battery according to the corrected SOC. If the SOC is not full according to the correction result, continue to replenish the battery of the cluster until it is full.

[0080] In this embodiment, the SOC correction module adopts the following algorithm:

[0081] SOC_corrected=SOC_predicted*f(Qc,η,k,...)

[0082] Wherein, f is a function that is corrected according to the eigenvalue.

[0083] S7, determining whether each battery cluster has been calibrated and fully charged;

[0084] In this embodiment, the inter-cluster switching and overall calibration are performed. When a cluster of batteries is fully charged, the current cluster is disconnected, and then the second cluster of batteries is connected and calibrated in the same way, and so on, until the SOC calibration and full charging of all cluster batteries are completed.

[0085] In this embodiment, the SOC calibration of the entire industrial and commercial energy storage compartment is completed based on the SOC calibration and full charge results of each battery cluster.

[0086] In summary, the present invention performs online SOC correction and calibration operations on single-cluster batteries of industrial and commercial energy storage systems. Each cluster of batteries is charged individually, and characteristic values ​​characterizing battery characteristics are obtained according to the charging data of the single-cluster batteries. A battery model including a BP neural network model is built, and then an estimated SOC is obtained through an SOC correction module according to the current charging status and characteristic values ​​of the background batteries, thereby realizing online correction of the SOC of the single-cluster battery, and replenishing the SOC of each cluster of batteries according to the correction results.

[0087] The present invention can obtain the SOC of the battery in real time through online correction, thereby improving the operating efficiency of the energy storage system. The present invention can be applied to industrial and commercial energy storage systems of different scales and configurations, and has strong versatility and scalability.

[0088] The present invention selects appropriate activation function and loss function, such as Sigmoid function and mean square error function, to improve the prediction accuracy of the model.

[0089] The present invention utilizes a large amount of historical charging data and corresponding SOC values ​​to train the BP neural network model, and adjusts the network weights and thresholds through the back propagation algorithm so that the model can accurately simulate the charging and discharging process of the battery.

[0090] The present invention utilizes the BP neural network model to simulate the battery charging and discharging process, combines the characteristic value to perform SOC correction, and improves the estimation accuracy of SOC.

[0091] The present invention corrects the prediction result of the BP neural network model through the SOC correction module according to the current charging status and characteristic value of the background battery, so as to obtain a more accurate SOC estimation value.

[0092] The present invention solves the technical problems of low SOC measurement accuracy, poor real-time performance and low charging and high discharging in the prior art.

[0093] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A battery SOC calibration method, characterized in that: The method comprises: S1. Maintain the normal operation of the battery pack, battery management system BMS and related equipment; obtain and control the test environment conditions based on the test environment information; S2, performing charging operation and data collection operation, performing constant current-constant voltage charging on each battery cluster in the battery pack individually, and during the charging operation, recording battery parameters in real time to form charging data; S3, performing a feature value extraction operation on the charging data to extract a feature value of a battery charging characteristic; S4. Building a battery model based on the characteristic value of the battery charging characteristic, wherein the battery model includes: a BP neural network model, allowing the BP neural network model to learn according to historical charging data to train and obtain a suitable prediction model, and using the suitable prediction model to simulate the charging and discharging process of the battery to predict the SOC value prediction result of the battery; S5, performing an SOC online correction operation, obtaining and correcting the SOC value prediction result through an SOC correction module according to the current charging status of the background battery and the characteristic value of the battery charging characteristic, and obtaining a corrected SOC estimated value; S6, performing a charging operation according to the corrected SOC estimated value until the battery of the current cluster reaches a fully charged state; S7, performing the correction operation of S5 and the power replenishment operation of S6 on the batteries of all the clusters in the battery pack until the correction operation and the power replenishment operation are completed on the battery pack.

2. A battery SOC calibration method according to claim 1, characterized in that: In S2, the battery parameters include: voltage V(t), current I(t) and temperature T(t).

3. A battery SOC calibration method according to claim 1, characterized in that: In S3, the battery charging characteristic values ​​include: charging capacity Qc, charging efficiency η and voltage curve slope k.

4. A battery SOC calibration method according to claim 1, characterized in that: In S3, the method for obtaining the characteristic value of the battery charging characteristic includes: polynomial fitting, spline function fitting, charging parameter analysis and voltage curve analysis.

5. A battery SOC calibration method according to claim 1, characterized in that: The S3 includes: S31. Calculate the charging capacity Qc received by the battery during the charging operation using the following logic: Qc=∫I(t)dt; S32. Calculate the charging efficiency η of the battery in the charging operation using the following logic: η = (final power - initial power) / (charged power)); S33. Calculate the voltage curve slope k using the following logic: k = dV(t) / dt.

6. A battery SOC calibration method according to claim 1, characterized in that: The S4 includes: S41, constructing the BP neural network model, wherein the BP neural network model includes: an input layer, a hidden layer and an output layer; S42, determining the number of layers and neurons of the hidden layer, and optimizing the network structure of the BP neural network model through preset test operations; S43, selecting an applicable activation function and an applicable loss function; S44, using the historical charging data and its corresponding SOC value to perform model training on the BP neural network model, adjusting the network weights and thresholds through a back propagation algorithm to obtain the applicable network model.

7. A battery SOC calibration method according to claim 1, characterized in that: In S41, the input layer is used to receive the battery charging characteristic feature value as input data, the input data is processed to obtain a predicted SOC value, and the output layer is used to output the SOC value prediction result.

8. A battery SOC calibration method according to claim 1, characterized in that: In S5, the SOC correction module uses the following logic to correct the SOC value prediction result: SOC_corrected=SOC_predicted*f(Qc,η,k,...) Where f is the function that is corrected according to the eigenvalue.

9. A battery SOC calibration method according to claim 1, characterized in that: The S7 includes: S71, performing inter-cluster switching operation and overall calibration operation on the battery group, and disconnecting the current cluster when the batteries in one cluster are fully charged; S72, executing S71 for the next group of batteries, connecting and calibrating, until the SOC calibration operation and full charging operation of the batteries in all groups are completed.

10. A battery SOC calibration system, characterized in that: The system comprises: Equipment and experimental condition setting module, used to maintain the normal operation of battery pack, battery management system BMS and related equipment; obtain and control the test environment conditions according to the test environment information; A charging and data acquisition module, used for performing charging and data acquisition operations, and performing constant current-constant voltage charging on each battery cluster in the battery pack. During the charging operation, battery parameters are recorded in real time to form charging data. The charging and data acquisition module is connected to the equipment and experimental condition setting module; A feature value extraction module, used for performing a feature value extraction operation on the charging data to extract a feature value of the battery charging characteristic, the feature value extraction module being connected to the charging and data acquisition module; A model building and training prediction module is used to build a battery model based on the characteristic value of the battery charging characteristic, wherein the battery model includes: a BP neural network model, which enables the BP neural network model to learn according to historical charging data to train and obtain a suitable prediction model, and uses the suitable prediction model to simulate the charging and discharging process of the battery to predict the SOC value prediction result of the battery, and the model building and training prediction module is connected to the characteristic value extraction module; An SOC correction module is used to perform an SOC online correction operation, obtain and correct the SOC value prediction result according to the current charging status of the background battery and the characteristic value of the battery charging characteristic through the SOC correction module to obtain a corrected SOC estimation value, and the SOC correction module is connected to the characteristic value extraction module; A charging module, used for performing a charging operation according to the corrected SOC estimated value until the battery of the current cluster reaches a fully charged state, the charging module being connected to the SOC correction module; The battery pack cyclic correction and power replenishment module is used to perform the correction operation of S5 and the power replenishment operation of S6 on the batteries of all clusters in the battery pack until the battery pack completes the correction operation and the power replenishment operation. The battery pack cyclic correction and power replenishment module is connected to the power replenishment module and the SOC correction module.

Images