A method for optimizing energy storage SOC balance control strategy

Abstract

The present invention discloses an energy storage SOC balance control strategy optimization method, which involves the field of machine learning, and includes: obtaining the battery service status; performing battery status prediction to obtain a set of SOH prediction values; performing open circuit voltage calibration in combination with the ambient temperature monitoring value to generate an open circuit voltage calibration value; based on the open circuit voltage calibration value and the ambient temperature monitoring value, in combination with the initial discharge current and the initial discharge cut-off voltage, traversing the SOH prediction value set to perform SOC prediction to obtain a set of SOC prediction values; performing balanced control optimization on the initial discharge current and the initial discharge cut-off voltage, generating a recommended discharge current and a recommended discharge cut-off voltage, and sending them to the user end. The method solves the technical problem of low control refinement in the existing energy storage SOC balance strategy, and achieves the technical effect of realizing refined balanced control of the battery pack SOC by accurately predicting and managing the SOH and SOC of a single battery cell.

Description

Technical Field

[0001] The present application relates to the field of machine learning, and in particular to a method for optimizing a storage SOC balancing control strategy. Background Art

[0002] In energy storage systems, the SOC (State of Charge, the remaining percentage of battery capacity) balance of the battery pack is crucial to the overall performance, life and safety of the battery pack. Unbalanced SOC will cause some batteries to be overcharged or over-discharged, shortening the battery life and even causing safety problems. Traditional battery health status assessment methods mainly focus on the entire battery system rather than a single cell, which leads to inaccurate assessment of battery system performance, low level of refinement of SOC balance control, and inability to effectively reduce performance differences between cells.

[0003] Among the current related technologies, the energy storage SOC balancing strategy has the technical problem of low degree of control refinement. Summary of the invention

[0004] The present application provides a method for optimizing an energy storage SOC balancing control strategy, which obtains the service status of the battery, accurately predicts the SOH of a single battery cell, calibrates the open circuit voltage in combination with the ambient temperature monitoring value, traverses the SOH prediction value set to predict the SOC, and based on the SOC prediction value set, performs balancing control and optimization of the initial discharge current and the initial discharge cut-off voltage, thereby achieving the technical effect of fine balancing control of the SOC of the battery pack by accurately predicting and managing the SOH and SOC of a single battery cell.

[0005] The present application provides a method for optimizing an energy storage SOC balancing control strategy, comprising:

[0006] Obtaining a battery service state, wherein the battery service state includes a charge and discharge record current sequence set, a charge and discharge cut-off voltage sequence, and an operating monitoring temperature sequence, wherein the charge and discharge record current sequence set corresponds one-to-one to a battery cell set, and the number of charge and discharge record current sequences corresponds to the number of charge and discharge cycles; predicting a battery state according to the charge and discharge record current sequence set, the charge and discharge cut-off voltage sequence, and the operating monitoring temperature sequence, and obtaining a SOH prediction value set, wherein the SOH prediction value set corresponds one-to-one to the battery cell set; performing open circuit voltage calibration according to the SOH prediction value set in combination with an ambient temperature monitoring value to generate an open circuit voltage calibration value; based on the open circuit voltage calibration value and the ambient temperature monitoring value, in combination with an initial discharge current and an initial discharge cut-off voltage, traversing the SOH prediction value set to perform SOC prediction and obtain an SOC prediction value set; based on the SOC prediction value set, performing balanced control optimization on the initial discharge current and the initial discharge cut-off voltage, generating a recommended discharge current and a recommended discharge cut-off voltage, and sending them to a user end.

[0007] In a possible implementation, the battery state is predicted according to the charge and discharge record current sequence set, the charge and discharge cut-off voltage sequence and the working monitoring temperature sequence to obtain a SOH prediction value set, and the following processing is performed:

[0008] Extract a first battery cell from the battery cell set; perform series-parallel analysis based on the charge and discharge record current sequence set in combination with the battery circuit topology to obtain a first battery cell charge and discharge record current sequence; perform series-parallel analysis based on the charge and discharge cut-off voltage sequence in combination with the battery circuit topology to obtain a first battery cell charge and discharge cut-off voltage sequence; predict the battery state of the first battery cell based on the first battery cell charge and discharge record current sequence, the first battery cell charge and discharge cut-off voltage sequence, and the working monitoring temperature sequence, obtain a first battery cell SOH prediction value, and add it to the SOH prediction value set.

[0009] In a possible implementation, the battery state of the first battery cell is predicted according to the first battery cell charge and discharge record current sequence, the first battery cell charge and discharge cut-off voltage sequence, and the working monitoring temperature sequence to obtain a first battery cell SOH prediction value, and the following processing is performed:

[0010] The first battery cell charge and discharge recording current sequence includes a first battery cell charge recording current sequence and a first battery cell discharge recording current sequence; the first battery cell charge and discharge cut-off voltage sequence includes a first battery cell charge cut-off voltage sequence and a first battery cell discharge cut-off voltage sequence; the working monitoring temperature sequence includes a charge monitoring temperature sequence and a discharge monitoring temperature sequence; through the SOH prediction network, the first battery cell charge recording current sequence, the first battery cell discharge recording current sequence, the first battery cell charge cut-off voltage sequence, the first battery cell discharge cut-off voltage sequence, the charge monitoring temperature sequence and the discharge monitoring temperature sequence are analyzed to generate the first battery cell SOH prediction value.

[0011] In a possible implementation, the first battery cell charging recorded current sequence, the first battery cell discharging recorded current sequence, the first battery cell charging cut-off voltage sequence, the first battery cell discharging cut-off voltage sequence, the charging monitoring temperature sequence and the discharging monitoring temperature sequence are parsed through the SOH prediction network to generate the first battery cell SOH prediction value, and the following processing is performed:

[0012] The first battery cell charge recording current sequence, the first battery cell discharge recording current sequence, the first battery cell charge cut-off voltage sequence, the first battery cell discharge cut-off voltage sequence, the charge monitoring temperature sequence and the discharge monitoring temperature sequence are segmented to obtain a first charge and discharge cycle array, wherein the first charge and discharge cycle array includes the first cycle charge recording current timing information, the first cycle charge cut-off voltage, the first cycle charge monitoring temperature, the first cycle discharge recording current timing information, the first cycle discharge cut-off voltage, and the first cycle discharge monitoring temperature in sequence; the first battery cell charge recording current sequence, the first battery cell discharge recording current sequence, the first battery cell charge cut-off voltage sequence, the first battery cell discharge cut-off voltage sequence The Nth charge and discharge cycle array is obtained by dividing the charge monitoring temperature sequence and the discharge monitoring temperature sequence, wherein the Nth charge and discharge cycle array includes the Nth cycle charge recording current timing information, the Nth cycle charge cut-off voltage, the Nth cycle charge monitoring temperature, the Nth cycle discharge recording current timing information, the Nth cycle discharge cut-off voltage, and the Nth cycle discharge monitoring temperature in sequence; an SOH prediction unit is obtained, wherein the SOH prediction unit includes a charging SOH prediction channel and a discharging SOH prediction channel connected in series; N SOH prediction units are copied in series to construct the SOH prediction network, the first charge and discharge cycle array is traversed until the Nth charge and discharge cycle array, and the first battery cell SOH prediction value is generated.

[0013] In a possible implementation, a SOH prediction unit is obtained and the following processing is performed:

[0014] Construct the SOH value prediction loss function:

[0015] Among them, LOSS represents the loss output value of any training, x 1i Characterizes the charging prediction value of the i-th attribute of SOH value, x 1i0 Characterizes the charging supervision value of the i-th attribute of the SOH value, x 2i Characterizes the discharge prediction value of the i-th attribute of SOH value, x 2i0 represents the discharge supervision value of the i-th attribute of the SOH value, M represents the number of SOH value attributes, and P i The i-th attribute deviation threshold is represented, c1 and c2 represent small constants to prevent the denominator from being equal to 0; according to the SOH value prediction loss function, the SOH prediction unit is trained based on the long short-term memory neural network.

[0016] In a possible implementation, the open circuit voltage is calibrated according to the SOH prediction value set and the ambient temperature monitoring value to generate an open circuit voltage calibration value, and the following processing is performed:

[0017] Performing mode analysis on the SOH prediction value set to obtain SOH characteristic values; collecting a historical open circuit voltage record data set with the same constraints based on the SOH characteristic values ​​in combination with the ambient temperature monitoring values; performing open circuit voltage calibration on the historical open circuit voltage record data set with the same constraints to generate the open circuit voltage calibration value.

[0018] In a possible implementation, based on the open circuit voltage calibration value and the ambient temperature monitoring value, combined with the initial discharge current and the initial discharge cut-off voltage, the SOH prediction value set is traversed to perform SOC prediction, and the SOC prediction value set is obtained, and the following processing is performed:

[0019] Collect a battery cell SOH value dataset, an open circuit voltage dataset, an ambient temperature dataset, a discharge current dataset, a discharge cut-off voltage dataset, and an SOC identification value dataset; configure a SOC prediction network based on the battery cell SOH value dataset, the open circuit voltage dataset, the ambient temperature dataset, the discharge current dataset, the discharge cut-off voltage dataset, and the SOC identification value dataset; through the SOC prediction network, based on the open circuit voltage calibration value, the ambient temperature monitoring value, the initial discharge current, and the initial discharge cut-off voltage, traverse the SOH prediction value set to perform SOC prediction and generate the SOC prediction value set.

[0020] In a possible implementation, based on the SOC prediction value set, the initial discharge current and the initial discharge cut-off voltage are balanced and optimized to generate a recommended discharge current and a recommended discharge cut-off voltage, and the following processing is performed:

[0021] A balanced fitness function is constructed, wherein the balanced fitness function is used to evaluate the variance value of the SOC prediction value set, and the balanced fitness function is provided with a balanced fitness threshold; a control fitness function is constructed, wherein the control fitness function is used to evaluate the mode value of the SOC prediction value set; and according to the balanced fitness function and the control fitness function, balanced control optimization is performed on the initial discharge current and the initial discharge cut-off voltage, and when the balanced fitness threshold is met and a preset number of optimization iterations are performed, the recommended discharge current and the recommended discharge cut-off voltage are output.

[0022] The energy storage SOC balancing control strategy optimization method proposed in this application is to first obtain the battery service state, wherein the battery service state includes a charge and discharge record current sequence set, a charge and discharge cut-off voltage sequence, and a working monitoring temperature sequence, wherein the charge and discharge record current sequence set corresponds to the battery cell set one by one, and the number of charge and discharge record current sequences corresponds to the number of charge and discharge cycles, and then the battery state is predicted according to the charge and discharge record current sequence set, the charge and discharge cut-off voltage sequence, and the working monitoring temperature sequence to obtain a SOH prediction value set, wherein the SOH prediction value set corresponds to the battery cell set one by one, and then according to the SOH prediction The open circuit voltage is calibrated in combination with the ambient temperature monitoring value to generate an open circuit voltage calibration value, and then based on the open circuit voltage calibration value and the ambient temperature monitoring value, combined with the initial discharge current and the initial discharge cut-off voltage, the SOH prediction value set is traversed to perform SOC prediction to obtain the SOC prediction value set, and finally, based on the SOC prediction value set, the initial discharge current and the initial discharge cut-off voltage are balanced and optimized to generate the recommended discharge current and the recommended discharge cut-off voltage, which are sent to the user end, thereby achieving the technical effect of realizing refined balanced control of the SOC of the battery pack by accurately predicting and managing the SOH and SOC of a single battery cell. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the accompanying drawings of the embodiment of the present invention will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the method according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0024] Figure 1 A flow chart of a method for optimizing an energy storage SOC balancing control strategy provided in an embodiment of the present application.

[0025] Figure 2A schematic diagram of a process for obtaining a set of SOH prediction values ​​in a method for optimizing an energy storage SOC balancing control strategy provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0027] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0028] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0029] The present application embodiment provides a method for optimizing energy storage SOC balance control strategy, such as Figure 1 As shown, the method includes:

[0030] Step S100, obtaining the battery service state, wherein the battery service state includes a charge and discharge record current sequence set, a charge and discharge cut-off voltage sequence, and a working monitoring temperature sequence, wherein the charge and discharge record current sequence set corresponds to the battery cell set one by one, and the number of charge and discharge record current sequences corresponds to the number of charge and discharge cycles. Specifically, the battery service state refers to the performance state of the battery during actual use, including its charge and discharge capacity, health status, temperature characteristics, etc.; the number of charge and discharge cycles refers to the number of times the battery completes a complete charge and discharge process (i.e., discharges from a fully charged state to an empty state, and then charges to a fully charged state). Specifically, a battery management system (BMS) or other data acquisition equipment is used to monitor and record the current data of the battery during the charge and discharge process in real time, and the charge and discharge current data of each cell (the basic unit of the battery, composed of a positive electrode, a negative electrode, an electrolyte, and a diaphragm, etc., is the smallest unit for the battery to perform a charge and discharge reaction) is organized in chronological order or charge and discharge cycle number to form a charge and discharge record current sequence set. Similarly, use BMS or data acquisition equipment to record the maximum and minimum voltage values ​​reached by the battery during the charge and discharge process, organize these voltage values ​​according to the number of charge and discharge cycles, and form a charge and discharge cut-off voltage sequence. Use temperature sensors to monitor the temperature of the battery in real time during operation, and organize these temperature data in chronological order or the number of charge and discharge cycles to form a working monitoring temperature sequence. That is, the charge and discharge recording current sequence set records the sequence of current changes of the battery during the charge and discharge process, which is used to analyze the charge and discharge characteristics and performance attenuation of the battery; the charge and discharge cut-off voltage sequence records the sequence of the maximum and minimum voltages reached by the battery during the charge and discharge process, which is used to judge the charge and discharge status and safety performance of the battery; the working monitoring temperature sequence records the sequence of temperature changes of the battery during operation, which is used to analyze the temperature characteristics and thermal management effects of the battery.

[0031] Step S200, predict the battery state according to the charge and discharge record current sequence set, the charge and discharge cut-off voltage sequence and the working monitoring temperature sequence, and obtain a set of SOH prediction values, wherein the set of SOH prediction values ​​corresponds one-to-one to the set of battery cells. Specifically, the charge and discharge record current sequence set, the charge and discharge cut-off voltage sequence and the working monitoring temperature sequence are cleaned, abnormal values ​​or missing values ​​are removed, and the data is normalized or standardized so that the model can better process the data. Extract key features from the charge and discharge record current sequence set, such as average charge current, average discharge current, current peak, etc.; extract features from the charge and discharge cut-off voltage sequence, such as maximum charge voltage, minimum discharge voltage, etc.; extract temperature-related features from the working monitoring temperature sequence, such as average temperature, maximum temperature, temperature fluctuation, etc. Use a prediction model (such as a neural network, support vector machine, random forest, etc.), take historical data, i.e., a known set of charge and discharge record current sequences, charge and discharge cut-off voltage sequences, and working monitoring temperature sequences as input, and the corresponding SOH value as output, to train the prediction model. During the training process, cross-validation and other techniques can be used to evaluate the performance of the prediction model, and hyperparameters can be adjusted to optimize the prediction model. Among them, the SOH (State of Health) refers to the health state of the battery, which indicates the percentage of the battery's current ability to store and release charge relative to its new state. Use the trained prediction model to predict the new set of charge and discharge record current sequences, charge and discharge cut-off voltage sequences, and working monitoring temperature sequences to obtain the corresponding SOH prediction value set. Each battery cell has a corresponding SOH prediction value, that is, the SOH prediction value set corresponds one-to-one to the battery cell set.

[0032] like Figure 2As shown, in a possible implementation, the battery state is predicted according to the charge and discharge record current sequence set, the charge and discharge cut-off voltage sequence and the working monitoring temperature sequence to obtain a SOH prediction value set, and step S200 further includes step S210, extracting the first battery cell from the battery cell set. Specifically, the battery cell set is traversed, and any one battery cell is selected as the processing object in sequence or according to a certain strategy, that is, the first battery cell. Step S220, according to the charge and discharge record current sequence set, combined with the battery circuit topology, a series-parallel analysis is performed to obtain the first battery cell charge and discharge record current sequence. Specifically, the charge and discharge record current sequence set of the entire battery pack is obtained, the circuit topology of the battery pack is analyzed (that is, the series-parallel relationship between the batteries), and the total current of the battery pack is distributed to each battery cell according to the circuit topology, so as to obtain the charge and discharge record current sequence of the first battery cell. Step S230, according to the charge and discharge cut-off voltage sequence, combined with the battery circuit topology, a series-parallel analysis is performed to obtain the first battery cell charge and discharge cut-off voltage sequence. Specifically, the charge and discharge cut-off voltage sequence of the entire battery pack is obtained, similar to step S220, and the total voltage of the battery pack is distributed to each battery cell according to the battery circuit topology to obtain the charge and discharge cut-off voltage sequence of the first battery cell. Step S240, based on the charge and discharge record current sequence of the first battery cell, the charge and discharge cut-off voltage sequence of the first battery cell, and the working monitoring temperature sequence, the battery state of the first battery cell is predicted, and the SOH prediction value of the first battery cell is obtained, and added to the SOH prediction value set. Specifically, using tools such as machine learning models, the SOH value of the first battery cell is predicted based on the charge and discharge record current sequence, charge and discharge cut-off voltage sequence, and working monitoring temperature sequence, and the predicted SOH value is added to the SOH prediction value set. This implementation method performs SOH prediction on each battery cell individually, achieving the technical effect of accurately evaluating the health status of each battery cell.

[0033] In a possible implementation, the battery state of the first battery cell is predicted according to the first battery cell charge and discharge record current sequence, the first battery cell charge and discharge cut-off voltage sequence, and the working monitoring temperature sequence to obtain a first battery cell SOH prediction value, and step S240 further includes step S241, the first battery cell charge and discharge record current sequence includes the first battery cell charge record current sequence and the first battery cell discharge record current sequence; step S242, the first battery cell charge and discharge cut-off voltage sequence includes the first battery cell charge cut-off voltage sequence and the first battery cell discharge cut-off voltage sequence; step S243, the working monitoring temperature sequence includes the charge monitoring temperature sequence and the discharge monitoring temperature sequence. Specifically, by marking and classifying the data, the first battery cell charge record current sequence and the first battery cell discharge record current sequence are distinguished from the first battery cell charge and discharge record current sequence, and similarly, the first battery cell charge cut-off voltage sequence and the first battery cell discharge cut-off voltage sequence are distinguished from the first battery cell charge and discharge cut-off voltage sequence, and the charge monitoring temperature sequence and the discharge monitoring temperature sequence are distinguished from the working monitoring temperature sequence. Step S244, through the SOH prediction network, the first battery cell charging record current sequence, the first battery cell discharging record current sequence, the first battery cell charging cut-off voltage sequence, the first battery cell discharging cut-off voltage sequence, the charging monitoring temperature sequence and the discharging monitoring temperature sequence are parsed to generate the first battery cell SOH prediction value. Specifically, using a pre-trained SOH prediction network (which can be a deep learning model, such as a neural network or a recurrent neural network), the first battery cell charging record current sequence, the first battery cell discharging record current sequence, the first battery cell charging cut-off voltage sequence, the first battery cell discharging cut-off voltage sequence, the charging monitoring temperature sequence and the discharging monitoring temperature sequence are used as inputs, and the SOH prediction value of the first battery cell is obtained through the forward propagation of the network. This implementation method subdivides the charging and discharging record current sequence, the charging and discharging cut-off voltage sequence and the working monitoring temperature sequence into two processes, charging and discharging, and more carefully analyzes the impact of the charging and discharging process on the battery health state, thereby achieving the technical effect of improving the prediction accuracy.

[0034] In one possible implementation, the first battery cell charge recording current sequence, the first battery cell discharge recording current sequence, the first battery cell charge cut-off voltage sequence, the first battery cell discharge cut-off voltage sequence, the charge monitoring temperature sequence and the discharge monitoring temperature sequence are parsed through the SOH prediction network to generate the first battery cell SOH prediction value, and step S244 further includes step S2441, which segments the first battery cell charge recording current sequence, the first battery cell discharge recording current sequence, the first battery cell charge cut-off voltage sequence, the first battery cell discharge cut-off voltage sequence, the charge monitoring temperature sequence and the discharge monitoring temperature sequence to obtain a first charge and discharge cycle array, wherein the first charge and discharge cycle array includes, in sequence, the first cycle charge recording current timing information, the first cycle charge cut-off voltage, the first cycle charge monitoring temperature, the first cycle discharge recording current timing information, the first cycle discharge cut-off voltage, and the first cycle discharge monitoring temperature. Specifically, the charge and discharge recording current sequence, the charge and discharge cut-off voltage sequence, and the charge and discharge monitoring temperature sequence of the first battery cell are segmented, and the segmentation is based on the charge and discharge cycle of the battery, that is, each complete charge and discharge process. After segmentation, a first charge and discharge cycle array is obtained, wherein the first charge and discharge cycle array contains the current timing information, cut-off voltage, and monitoring temperature of a complete charge and discharge cycle. Step S2442, the first battery cell charge recording current sequence, the first battery cell discharge recording current sequence, the first battery cell charge cut-off voltage sequence, the first battery cell discharge cut-off voltage sequence, the charge monitoring temperature sequence, and the discharge monitoring temperature sequence are segmented to obtain the Nth charge and discharge cycle array, wherein the Nth charge and discharge cycle array includes the Nth cycle charge recording current timing information, the Nth cycle charge cut-off voltage, the Nth cycle charge monitoring temperature, the Nth cycle discharge recording current timing information, the Nth cycle discharge cut-off voltage, and the Nth cycle discharge monitoring temperature. Specifically, similarly, continue to divide the data until the Nth charge and discharge cycle array is obtained, and the Nth charge and discharge cycle array also contains the current timing information, cut-off voltage and monitoring temperature of a complete charge and discharge cycle, but these data correspond to the Nth charge and discharge cycle. Step S2443, obtain the SOH prediction unit, wherein the SOH prediction unit includes a charging SOH prediction channel and a discharging SOH prediction channel connected in series. Specifically, design a SOH prediction unit, comprising two channels connected in series: a charging SOH prediction channel and a discharging SOH prediction channel, wherein the charging SOH prediction channel receives charging data (charging current timing information, charging cut-off voltage and charging monitoring temperature) to predict the SOH value; the discharging SOH prediction channel further predicts the SOH value based on the charging prediction and the discharging data (discharging current timing information, discharging cut-off voltage and discharging monitoring temperature), thereby obtaining the SOH value at the end of a charge and discharge cycle.Step S2444, copy N of the SOH prediction units and connect them in series to construct the SOH prediction network, traverse the first charge and discharge cycle array until the Nth charge and discharge cycle array, and generate the SOH prediction value of the first battery cell. Specifically, copy N SOH prediction units, and connect the N SOH prediction units in series to construct a complete SOH prediction network. Use this SOH prediction network to traverse all data from the first charge and discharge cycle array to the Nth charge and discharge cycle array. For each charge and discharge cycle array, first make a preliminary prediction through the charging SOH prediction channel, and then make a final prediction in combination with the discharging SOH prediction channel. Finally, combine the SOH prediction values ​​of each charge and discharge cycle to obtain the overall SOH prediction value of the first battery cell. This implementation method accurately analyzes the impact of each cycle on the battery health status by dividing the charge and discharge data and temperature data according to the charge and discharge cycle, thereby achieving the technical effect of improving the accuracy of SOH prediction.

[0035] In a possible implementation, a SOH prediction unit is obtained, and step S2443 further includes step S24431, constructing a SOH value prediction loss function:

[0036]

[0037] Among them, LOSS represents the loss output value of any training, x 1i Characterizes the charging prediction value of the i-th attribute of SOH value, x 1i0 Characterizes the charging supervision value of the i-th attribute of the SOH value, x 2i Characterizes the discharge prediction value of the i-th attribute of SOH value, x 2i0 represents the discharge supervision value of the i-th attribute of the SOH value, M represents the number of SOH value attributes, and P i Characterize the deviation threshold of the i-th attribute, c1 and c2 characterize small constants to prevent the denominator from being equal to 0; step S24432, according to the SOH value prediction loss function, based on the long short-term memory neural network, train the SOH prediction unit. Specifically, select the long short-term memory neural network (LSTM) as the basic model of the SOH prediction unit, prepare training data, including battery charging and discharging records, cut-off voltage, monitoring temperature and other time series data, and the corresponding SOH supervision value, use the loss function defined in step S24431 as the training target, and optimize the model parameters through the back propagation algorithm to make the predicted value gradually approach the true value. This implementation method trains the model by minimizing the loss function, gradually optimizes the model parameters, and improves the prediction ability of the model. The constructed loss function can amplify the loss value deviation of different model parameters, achieving the technical effect of facilitating accurate adjustment and rapid convergence of model parameters.

[0038] Step S300, according to the SOH prediction value set, the open circuit voltage is calibrated in combination with the ambient temperature monitoring value to generate an open circuit voltage calibration value. Specifically, the open circuit voltage refers to the voltage difference between the positive and negative electrodes of the battery when no current passes through the battery (i.e., open circuit state). Specifically, the SOH prediction value set includes the SOH prediction value of each battery cell, and the current ambient temperature monitoring value is obtained through the ambient temperature sensor. According to the existing experimental data or historical data, the relationship between the open circuit voltage, SOH and ambient temperature is analyzed, and a relationship model between the open circuit voltage, SOH and ambient temperature is established through data fitting or machine learning methods. The SOH prediction value set and the ambient temperature monitoring value are substituted into the established relationship model, and the open circuit voltage calibration value at the current SOH and ambient temperature is obtained by calculation.

[0039] In a possible implementation, according to the SOH prediction value set, the open circuit voltage is calibrated in combination with the ambient temperature monitoring value to generate the open circuit voltage calibration value, and step S300 further includes step S310, performing mode analysis according to the SOH prediction value set to obtain the SOH characteristic value. Specifically, the SOH prediction value set is subjected to mode analysis, and the mode is the value that appears the most times in a set of data, representing the "center point" or "typical value" of this set of data, and the mode is selected as the SOH characteristic value. Step S320, based on the SOH characteristic value and in combination with the ambient temperature monitoring value, a data set of historical open circuit voltage records with the same constraints is collected. Specifically, according to the SOH characteristic value and the ambient temperature monitoring value, the open circuit voltage records matching these two conditions are retrieved from the historical database, and these records are obtained under conditions similar to the current conditions, and the retrieved records are formed into a data set, which contains historical open circuit voltage records similar to the current conditions. Step S330, the open circuit voltage is calibrated on the data set of historical open circuit voltage records with the same constraints to generate the open circuit voltage calibration value. Specifically, statistical analysis is performed on the same constraint historical open circuit voltage record data set, including the average value, median or other appropriate statistical indicators, and a statistical indicator (such as the average value) is selected as the calibration value of the open circuit voltage, which represents the open circuit voltage of the battery under given SOH and ambient temperature conditions. This implementation method uses historical data similar to the current conditions to calibrate the open circuit voltage, achieving the technical effect of improving the accuracy and reliability of the calibration.

[0040] Step S400, based on the open circuit voltage calibration value and the ambient temperature monitoring value, combined with the initial discharge current and the initial discharge cut-off voltage, traverse the SOH prediction value set to perform SOC prediction and obtain the SOC prediction value set. Specifically, determine the initial discharge current and initial discharge cut-off voltage of the battery, wherein the initial discharge current is the current value when the battery starts to discharge; the initial discharge cut-off voltage is the voltage value when the battery is discharged to the end. Based on the working principle and experimental data of the battery, analyze the relationship between SOC (State of Charge, the percentage of remaining power, which indicates the ratio of the amount of charge that the battery can currently provide to the amount of charge that it can provide when fully charged) and open circuit voltage, ambient temperature, discharge current and discharge cut-off voltage, and establish a relationship model between SOC and open circuit voltage, ambient temperature, discharge current and discharge cut-off voltage through data fitting, machine learning or table lookup. For each SOH prediction value in the SOH prediction value set, the current ambient temperature monitoring value, initial discharge current and initial discharge cut-off voltage, and open circuit voltage calibration value are substituted into the established relationship model, and the corresponding SOC prediction value is obtained by calculation. All the obtained SOC prediction values ​​are combined into an SOC prediction value set, and the SOC prediction value set corresponds to the battery cell set one by one.

[0041] In a possible implementation, based on the open circuit voltage calibration value and the ambient temperature monitoring value, combined with the initial discharge current and the initial discharge cut-off voltage, the SOH prediction value set is traversed to perform SOC prediction to obtain the SOC prediction value set, and step S400 further includes step S410, collecting the battery cell SOH value data set, the open circuit voltage data set, the ambient temperature data set, the discharge current data set, the discharge cut-off voltage data set and the SOC identification value data set. Specifically, historical battery cell SOH values, historical ambient temperature data, and open circuit voltage data at different battery cell SOH values ​​and ambient temperatures are collected from the battery management system, initial discharge current data and discharge cut-off voltage data are collected, and the SOC actual value corresponding to the above data is obtained as the SOC identification value. Step S420, based on the battery cell SOH value data set, the open circuit voltage data set, the ambient temperature data set, the discharge current data set, the discharge cut-off voltage data set and the SOC identification value data set, configure the SOC prediction network. Specifically, the collected data set is cleaned, standardized or normalized, etc., to improve the quality of the data and the efficiency of network training. Based on the neural network architecture, the preprocessed data set is used as input, and the SOC identification value is used as the target output. The SOC prediction network is trained to learn the mapping relationship from input to output. During the training process, the network hyperparameters, such as learning rate, batch size, number of iterations, etc., need to be adjusted to optimize the performance of the network. Step S430, through the SOC prediction network, based on the open circuit voltage calibration value, the ambient temperature monitoring value, the initial discharge current, and the initial discharge cut-off voltage, the SOH prediction value set is traversed to perform SOC prediction and generate the SOC prediction value set. Specifically, the input data to be predicted, including the open circuit voltage calibration value, the ambient temperature monitoring value, the initial discharge current and the initial discharge cut-off voltage, and each SOH prediction value, are used as the input of the SOC prediction network. The SOC prediction network outputs the corresponding SOC prediction value according to the input data, and the SOC prediction values ​​corresponding to each SOH prediction value are combined into a set as the final SOC prediction result. This implementation method achieves the technical effect of improving the accuracy and reliability of SOC prediction by configuring the SOC prediction network.

[0042] Step S500, based on the SOC prediction value set, the initial discharge current and the initial discharge cut-off voltage are optimized by balanced control, and a recommended discharge current and a recommended discharge cut-off voltage are generated and sent to the user end. Specifically, the SOC distribution of the battery pack is analyzed, and the SOC state of each battery cell is determined according to the SOC prediction value set. According to the balanced control strategy of the battery pack, it is determined how to adjust the discharge current and the discharge cut-off voltage to achieve SOC balance between the batteries, wherein the balanced control is to balance the SOC between the batteries by adjusting the discharge parameters to extend the service life and improve the performance of the battery pack. Based on the initial discharge current and the initial discharge cut-off voltage of the current battery, and the SOC prediction value set, an optimization calculation is performed through an algorithm or model. The goal of the optimization is to find a set of recommended discharge currents and recommended discharge cut-off voltages so that the battery pack can achieve the best SOC balancing effect during the discharge process. At the same time, combined with factors such as battery safety, efficiency and life, the optimization algorithm may include a numerical optimization algorithm (such as a gradient descent method, a genetic algorithm, etc.) or a rule-based logical judgment. According to the results of the optimization calculation, a recommended discharge current and a recommended discharge cut-off voltage are generated, and these recommended values ​​are used as new discharge parameters to guide the discharge process of the battery pack. The generated recommended discharge current and recommended discharge cut-off voltage are sent to the user end through communication methods such as Bluetooth, Wi-Fi, CAN bus, etc. The user end can be a battery management system, a central control unit, or other equipment responsible for battery monitoring and control. The embodiment of the present application adopts the technical means of obtaining the service status of the battery, accurately predicting the SOH of a single battery cell, calibrating the open circuit voltage in combination with the ambient temperature monitoring value, traversing the SOH prediction value set to predict the SOC, and based on the SOC prediction value set, performing balanced control optimization on the initial discharge current and the initial discharge cut-off voltage, etc., to achieve the technical effect of fine balanced control of the SOC of the battery pack by accurately predicting and managing the SOH and SOC of a single battery cell.

[0043] In a possible implementation, based on the SOC prediction value set, the initial discharge current and the initial discharge cut-off voltage are balanced and optimized to generate a recommended discharge current and a recommended discharge cut-off voltage. Step S500 further includes step S510, constructing a balanced fitness function, wherein the balanced fitness function is used to evaluate the variance value of the SOC prediction value set, and the balanced fitness function is provided with a balanced fitness threshold. Specifically, according to the SOC prediction value set, its variance value is calculated, and the variance value reflects the discrete degree of data in the SOC prediction value set, that is, the fluctuation of the SOC prediction value. Using the variance value as input, a function is defined to measure the balance of the SOC prediction value. For example, the balanced fitness function can be designed to be inversely proportional to the variance value, that is, the smaller the variance value, the higher the balanced fitness. According to actual needs, a balanced fitness threshold is set to determine whether the balanced control of the discharge current and the discharge cut-off voltage meets the requirements. Step S520, constructing a control fitness function, wherein the control fitness function is used to evaluate the mode value of the SOC prediction value set. Specifically, in the SOC prediction value set, find the SOC value that appears the most times, that is, the mode value, which represents the most likely SOC state. Use the mode value as input to define a function to measure the control performance of the SOC prediction value. The control fitness function can be designed as a function related to the mode value, such as the closer the mode value is to a target value, the higher the control fitness. Step S530, according to the balanced fitness function and the control fitness function, the initial discharge current and the initial discharge cut-off voltage are balanced and optimized, and when the balanced fitness threshold is met and the preset number of optimization iterations are performed, the recommended discharge current and the recommended discharge cut-off voltage are output. Specifically, the search range of the initial discharge current and the initial discharge cut-off voltage, as well as the parameters of the optimization algorithm (such as the number of iterations, learning rate, etc.) are set. Within the search range, the values ​​of the discharge current and the discharge cut-off voltage are continuously adjusted, the corresponding SOC prediction value set is calculated, and the balance and control performance under the current discharge current and discharge cut-off voltage are evaluated according to the equilibrium fitness function and the control fitness function, respectively. Among them, the goal of the control fitness function is to find a suitable combination of discharge current and discharge cut-off voltage so that the mode value of the SOC prediction value is close to the expected target value and meets the equilibrium fitness threshold. If the equilibrium fitness under the current parameters meets the equilibrium fitness threshold requirements and reaches the preset number of iterative optimization times, the iteration is stopped and the current discharge current and discharge cut-off voltage are output as recommended values; if the stop condition is not met, the values ​​of the discharge current and discharge cut-off voltage are updated according to the optimization algorithm (such as gradient descent method, genetic algorithm, etc.), and the iterative optimization continues.This implementation method constructs a balance fitness function and a control fitness function so that the battery's SOC prediction value set can have good control performance while satisfying a certain balance, achieving the technical effect of enabling the output recommended discharge current and recommended discharge cut-off voltage to enable the battery's SOC prediction value set to achieve the best technical effect in terms of balance and control performance.

[0044] The above specific implementation manner does not constitute a limitation to the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps recorded in the present application can be performed in an order different from that in the embodiment and can still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for optimizing energy storage SOC balance control strategy, characterized in that: include: Obtaining a battery service state, wherein the battery service state includes a charge and discharge record current sequence set, a charge and discharge cut-off voltage sequence, and an operating monitoring temperature sequence, wherein the charge and discharge record current sequence set corresponds to a battery cell set one by one, and the number of charge and discharge record current sequences corresponds to the number of charge and discharge cycles; Predicting the battery state according to the charge and discharge record current sequence set, the charge and discharge cut-off voltage sequence and the working monitoring temperature sequence to obtain a SOH prediction value set, wherein the SOH prediction value set corresponds one-to-one to the battery cell set; According to the SOH prediction value set, open circuit voltage calibration is performed in combination with the ambient temperature monitoring value to generate an open circuit voltage calibration value; Based on the open circuit voltage calibration value and the ambient temperature monitoring value, combined with the initial discharge current and the initial discharge cut-off voltage, traverse the SOH prediction value set to perform SOC prediction to obtain an SOC prediction value set; Based on the SOC prediction value set, the initial discharge current and the initial discharge cut-off voltage are balanced and optimized to generate a recommended discharge current and a recommended discharge cut-off voltage, and send them to a user end; The battery state prediction is performed according to the charge and discharge record current sequence set, the charge and discharge cut-off voltage sequence and the working monitoring temperature sequence to obtain the SOH prediction value set, including: Extracting a first battery cell from the battery cell collection; According to the charge and discharge record current sequence set, a series-parallel analysis is performed in combination with the battery circuit topology to obtain a first battery cell charge and discharge record current sequence; According to the charge and discharge cut-off voltage sequence, a series-parallel analysis is performed in combination with the battery circuit topology to obtain a first battery cell charge and discharge cut-off voltage sequence; According to the first battery cell charge and discharge record current sequence, the first battery cell charge and discharge cut-off voltage sequence and the working monitoring temperature sequence, the first battery cell battery state prediction is performed to obtain a first battery cell SOH prediction value, and the SOH prediction value is added to the SOH prediction value set; According to the first battery cell charge and discharge record current sequence, the first battery cell charge and discharge cut-off voltage sequence and the working monitoring temperature sequence, the battery state of the first battery cell is predicted to obtain the first battery cell SOH prediction value, including: The first battery cell charge and discharge recording current sequence includes a first battery cell charge recording current sequence and a first battery cell discharge recording current sequence; The first battery cell charge and discharge cut-off voltage sequence includes a first battery cell charge cut-off voltage sequence and a first battery cell discharge cut-off voltage sequence; The working monitoring temperature sequence includes a charging monitoring temperature sequence and a discharging monitoring temperature sequence; The first battery cell charge recorded current sequence, the first battery cell discharge recorded current sequence, the first battery cell charge cut-off voltage sequence, the first battery cell discharge cut-off voltage sequence, the charge monitoring temperature sequence and the discharge monitoring temperature sequence are parsed through the SOH prediction network to generate the first battery cell SOH prediction value; The first battery cell charge record current sequence, the first battery cell discharge record current sequence, the first battery cell charge cut-off voltage sequence, the first battery cell discharge cut-off voltage sequence, the charge monitoring temperature sequence and the discharge monitoring temperature sequence are parsed through the SOH prediction network to generate the first battery cell SOH prediction value, including: The first battery cell charge recording current sequence, the first battery cell discharge recording current sequence, the first battery cell charge cut-off voltage sequence, the first battery cell discharge cut-off voltage sequence, the charge monitoring temperature sequence and the discharge monitoring temperature sequence are segmented to obtain a first charge and discharge cycle array, wherein the first charge and discharge cycle array sequentially includes first cycle charge recording current timing information, first cycle charge cut-off voltage, first cycle charge monitoring temperature, first cycle discharge recording current timing information, first cycle discharge cut-off voltage and first cycle discharge monitoring temperature; The first battery cell charge recording current sequence, the first battery cell discharge recording current sequence, the first battery cell charge cut-off voltage sequence, the first battery cell discharge cut-off voltage sequence, the charge monitoring temperature sequence and the discharge monitoring temperature sequence are segmented to obtain an Nth charge and discharge cycle array, wherein the Nth charge and discharge cycle array sequentially includes the Nth cycle charge recording current timing information, the Nth cycle charge cut-off voltage, the Nth cycle charge monitoring temperature, the Nth cycle discharge recording current timing information, the Nth cycle discharge cut-off voltage and the Nth cycle discharge monitoring temperature; Obtaining a SOH prediction unit, wherein the SOH prediction unit includes a charging SOH prediction channel and a discharging SOH prediction channel connected in series; Duplicate N of the SOH prediction units and connect them in series to construct the SOH prediction network, traverse the first charge-discharge cycle array until the Nth charge-discharge cycle array, and generate the first battery cell SOH prediction value; The SOH prediction unit is obtained, including: Construct the SOH value prediction loss function: ， in, Represents the loss output value of any training, Characterizes the charging prediction value of the i-th attribute of the SOH value, Characterizes the charging supervision value of the i-th attribute of the SOH value, Characterizes the discharge prediction value of the i-th attribute of the SOH value, represents the discharge supervision value of the i-th attribute of the SOH value, M represents the number of SOH value attributes, Characterizes the deviation threshold of the i-th attribute, and Represents a small constant to prevent the denominator from being equal to 0; According to the SOH value prediction loss function, based on a long short-term memory neural network, training the SOH prediction unit; The open circuit voltage is calibrated according to the SOH prediction value set and combined with the ambient temperature monitoring value to generate the open circuit voltage calibration value, including: Performing a mode analysis on the SOH prediction value set to obtain a SOH characteristic value; Based on the SOH characteristic value and in combination with the ambient temperature monitoring value, a historical open circuit voltage record data set with the same constraint is collected; Performing open circuit voltage calibration on the same-constraint historical open circuit voltage record data set to generate the open circuit voltage calibration value; Wherein, based on the open circuit voltage calibration value and the ambient temperature monitoring value, combined with the initial discharge current and the initial discharge cut-off voltage, the SOH prediction value set is traversed to perform SOC prediction, and the SOC prediction value set is obtained, including: Collect the battery cell SOH value data set, open circuit voltage data set, ambient temperature data set, discharge current data set, discharge cut-off voltage data set and SOC identification value data set; Based on the battery cell SOH value data set, the open circuit voltage data set, the ambient temperature data set, the discharge current data set, the discharge cut-off voltage data set and the SOC identification value data set, configuring an SOC prediction network; Through the SOC prediction network, based on the open circuit voltage calibration value, the ambient temperature monitoring value, the initial discharge current and the initial discharge cut-off voltage, the SOH prediction value set is traversed to perform SOC prediction and generate the SOC prediction value set; The method of performing balanced control optimization on the initial discharge current and the initial discharge cut-off voltage based on the SOC prediction value set to generate a recommended discharge current and a recommended discharge cut-off voltage includes: Constructing a balanced fitness function, wherein the balanced fitness function is used to evaluate the variance value of the SOC prediction value set, and the balanced fitness function is provided with a balanced fitness threshold; Constructing a control fitness function, wherein the control fitness function is used to evaluate the mode value of the SOC prediction value set; According to the equilibrium fitness function and the control fitness function, the initial discharge current and the initial discharge cut-off voltage are subjected to equilibrium control optimization, and when the equilibrium fitness threshold is met and a preset number of optimization iterations are performed, the recommended discharge current and the recommended discharge cut-off voltage are output.

Images