Battery BMS (Battery Management System) management algorithm optimal feature engineering method and device

Abstract

The embodiment of the invention provides a battery BMS management algorithm optimization feature engineering method and device, and the method comprises the steps: collecting the historical data of each single battery in each battery cluster in a power station, and constructing sample data at equal time intervals based on the historical data; on the basis of the sample data at the equal time intervals, sequentially obtaining feature data including power utilization state feature data, power utilization state segment feature data and single-time power utilization accumulated charge and discharge capacity feature data; based on electricity utilization state characteristic data in the characteristic data, the electricity utilization state segment characteristic data and the single-time electricity utilization accumulated charge and discharge capacity characteristic data, respectively constructing label data comprising a single battery SOC label, a single battery internal resistance label and a single battery SOH label; and performing BMS management algorithm feature analysis on the historical data, the feature data and the tag data to obtain an optimal feature combination of the BMS management algorithm.

Description

Technical Field

[0001] The present application relates to the technical field of power system management, and in particular to a method and device for optimizing feature engineering of a battery BMS management algorithm. Background Art

[0002] The energy storage battery BMS management system implements the following functions through algorithms:

[0003] (1) Energy balance management between single cells: including battery consistency monitoring such as voltage consistency, internal resistance consistency, and SOC consistency. During the battery charging and discharging process, each battery will have performance differences such as internal resistance and capacity. If not monitored and managed, the differences will gradually increase with the increase in the number of charge and discharge cycles, causing some single cells to be overcharged or over-discharged, thereby affecting the performance and life of the entire battery pack.

[0004] (2) Internal resistance estimation: Accurate estimation of internal resistance can detect deteriorated batteries in a timely manner to prevent individual batteries from affecting the performance and remaining life of the entire battery pack.

[0005] (3) SOC estimation: Accurately estimate the state of charge (SOC) of the power battery pack, that is, the remaining battery power, to ensure that the SOC is maintained within a reasonable range and prevent damage to the battery due to overcharging or over-discharging.

[0006] (4) SOH estimation: Accurately estimate SOH to prevent potential safety issues caused by abnormal battery aging.

[0007] Feature engineering: It is generally believed to include three parts: feature construction, feature extraction, and feature selection. Feature engineering converts data into features that better represent potential problems, thereby improving the learning performance of the algorithm. The better the features, the better the model performance and the greater the flexibility. However, the existing technology has the difficulty that the original data information is not conducive to directly calculating the battery internal resistance, SOC, SOH and other battery BMS indicators, and it is also impossible to provide high-quality and effective data support for the internal resistance algorithm, SOC algorithm, SOH algorithm, etc. related to the energy storage battery BMS management. Summary of the invention

[0008] In view of this, an embodiment of the present application provides a method for optimizing feature engineering of a battery BMS management algorithm. One or more embodiments of the present application also involve a device for optimizing feature engineering of a battery BMS management algorithm, a computing device, a computer-readable storage medium, and a computer program to solve the technical defects existing in the prior art.

[0009] According to a first aspect of an embodiment of the present application, a method for optimizing feature engineering of a battery BMS management algorithm is provided, comprising:

[0010] Collect historical data including cell voltage, cell current and cell temperature of each cell in each battery cluster in the power station, and construct sample data with equal time intervals based on the historical data;

[0011] Based on the sample data at equal time intervals, characteristic data including power usage status characteristic data, power usage status segment characteristic data and single power usage cumulative charge and discharge capacity characteristic data are obtained in sequence;

[0012] Based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label are constructed respectively;

[0013] By performing a BMS management algorithm feature analysis on the historical data, the feature data and the tag data, an optimal feature combination of the BMS management algorithm is obtained.

[0014] Preferably, the characteristic data including power consumption state characteristic data, power consumption state segment characteristic data and single power consumption cumulative charge and discharge capacity characteristic data obtained in sequence based on the sample data at equal time intervals include:

[0015] Determine the power consumption status characteristic data of each single battery based on the single battery current in the sample data at equal time intervals;

[0016] According to the power usage state characteristic data of each single battery, the same continuous power usage state characteristic data is divided into a power usage state segment characteristic data, and a power usage state segment number is set for each power usage state segment characteristic data in time order;

[0017] The single-time power consumption cumulative charge-discharge capacity characteristic data of each single battery is obtained by calculating the cumulative charge-discharge capacity of the power consumption status segment characteristic data corresponding to each power consumption status segment number.

[0018] Preferably, the label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label are constructed based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, respectively, and include:

[0019] Based on the charge and discharge SOC in the SOC-OCV table and the voltage value and rated capacity corresponding to each SOC value, convert the charge and discharge SOC into a rated cumulative power consumption capacity, and construct a new SOC-OCV table based on the rated cumulative power consumption capacity;

[0020] Based on the power usage state characteristic data of each single battery and the power usage state segment characteristic data, extracting a first power usage state segment in which the previous power usage state is a static state and the static time length is greater than a preset time length;

[0021] Taking the voltage at the last moment of the previous power usage state as the starting voltage of the first power usage state segment, and determining the rated cumulative power usage capacity corresponding to the starting voltage of the first power usage state segment based on the new SOC-OCV table;

[0022] The rated cumulative power consumption capacity is used as the starting cumulative charging capacity of the first power consumption state segment, and the single cell SOC label is obtained based on the single power consumption cumulative charging and discharging capacity characteristic data of each single cell, the starting cumulative charging capacity of power consumption and the rated capacity data of the single cell.

[0023] Preferably, the label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label are constructed based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, respectively, and include:

[0024] Based on the new SOC-OCV table and the single-time power consumption cumulative charge and discharge capacity characteristic data of each single battery, find out the single battery voltage and OCV corresponding to the cumulative power consumption capacity of each power consumption record;

[0025] The internal resistance label of the single cell battery is obtained based on the single cell voltage, OCV and the single cell current corresponding to the cumulative power consumption capacity of each section of the power consumption record.

[0026] Preferably, the label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label are constructed based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, respectively, and include:

[0027] Based on the power usage status segmentation feature data, determine each power usage status level, and extract the power usage status segments of the full charging period level and the previous period and the next period being both the deep static period level from all the power usage status levels;

[0028] Obtaining the extracted starting open circuit voltage and ending open circuit voltage of the power usage state segment, and obtaining the starting cumulative charging capacity and ending cumulative charging capacity corresponding to the starting open circuit voltage to the ending open circuit voltage based on the new SOC-OCV table;

[0029] Based on the starting cumulative charging capacity and the ending cumulative charging capacity, the rated cumulative charging capacity change under the rated capacity condition is calculated, and based on the rated cumulative charging capacity change and the cumulative charging capacity, the single cell SOH label is obtained.

[0030] Preferably, the performing BMS management algorithm feature analysis on the historical data, the feature data and the tag data to obtain the optimal feature combination of the BMS management algorithm includes:

[0031] Based on the correlation analysis algorithm, any one of the single cell SOC label, the single cell internal resistance label and the single cell SOH label is used as label data, and any two of them are used as feature data, and the historical data and the feature data are used as feature data, and the correlation coefficient of each feature data and the label data is calculated respectively;

[0032] Based on the random forest algorithm, any one of the single cell SOC label, the single cell internal resistance label and the single cell SOH label is used as label data, and any two of them are used as feature data, and the historical data and the feature data are used as feature data, and the importance index of each feature data and label data is calculated respectively;

[0033] Based on the correlation coefficient and the importance index, the comprehensive correlation index of each feature data and label data is calculated respectively, and the feature data and label data with effective comprehensive correlation index are selected;

[0034] Based on the long short-term memory artificial neural network algorithm and the feature data and label data corresponding to the effective comprehensive correlation index, the optimal feature combination of the BMS management algorithm is obtained.

[0035] Preferably, the feature data and label data corresponding to the effective comprehensive correlation index based on the long short-term memory artificial neural network algorithm to obtain the optimal feature combination of the BMS management algorithm includes:

[0036] When the label data is a single battery SOC label, based on the long short-term memory artificial neural network algorithm and the selected effective comprehensive correlation index feature data and the single battery SOC label, a feature combination with the smallest SOC model error is calculated, and the feature combination is used as the optimal feature combination of the SOC model;

[0037] When the label data is a single cell internal resistance label, based on the long short-term memory artificial neural network algorithm and the selected effective comprehensive correlation index feature data and the single cell internal resistance label, a feature combination with the minimum error of the internal resistance model is calculated, and the feature combination is used as the optimal feature combination of the internal resistance model;

[0038] When the label data is a single cell SOH label, based on the long short-term memory artificial neural network algorithm and the selected effective comprehensive correlation index feature data and the single cell SOH label, the feature combination with the smallest SOH model error is calculated, and the feature combination is used as the optimal feature combination of the SOH model.

[0039] According to a second aspect of an embodiment of the present application, a battery BMS management algorithm optimization feature engineering device is provided, comprising:

[0040] A collection module is configured to collect historical data including cell voltage, cell current and cell temperature of each cell in each battery cluster in the power station, and construct sample data with equal time intervals based on the historical data;

[0041] The characteristic data acquisition module is configured to sequentially obtain characteristic data including power consumption state characteristic data, power consumption state segment characteristic data and single power consumption cumulative charge and discharge capacity characteristic data based on the sample data at equal time intervals;

[0042] A label data acquisition module is configured to construct label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label respectively based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data;

[0043] The module for obtaining the optimal feature combination is configured to obtain the optimal feature combination of the BMS management algorithm by performing a BMS management algorithm feature analysis on the historical data, the feature data and the tag data.

[0044] According to a third aspect of an embodiment of the present application, a computing device is provided, including:

[0045] Memory and processor;

[0046] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement any step of the battery BMS management algorithm optimization feature engineering method.

[0047] According to a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, which stores computer-executable instructions, and when the instructions are executed by a processor, the steps of any one of the battery BMS management algorithm optimization feature engineering methods are implemented.

[0048] According to a fifth aspect of an embodiment of the present application, a computer program is provided, wherein when the computer program is executed in a computer, the computer is caused to execute the steps of the above-mentioned battery BMS management algorithm optimization feature engineering method.

[0049] The preferred feature engineering scheme for the battery BMS management algorithm provided in the embodiment of the present application collects historical data of each single cell in each battery cluster in the power station, and constructs sample data with equal time intervals based on the historical data; based on the sample data with equal time intervals, characteristic data including power consumption status characteristic data, power consumption status segmentation characteristic data and single power consumption cumulative charge and discharge capacity characteristic data are obtained in sequence; based on the power consumption status characteristic data, the power consumption status segmentation characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, label data including single cell SOC label, single cell internal resistance label and single cell SOH label are constructed respectively; by performing BMS management algorithm feature analysis on the historical data, the characteristic data and the label data, the optimal feature combination of the BMS management algorithm is obtained. The preferred feature engineering scheme for the battery BMS management algorithm provided in the embodiment of the present application constructs the features required by the BMS management related algorithms (internal resistance, SOC, SOH) by preprocessing the data for the battery, and satisfies the data support for the algorithm development to efficiently output the optimal results. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a flow chart of a method for optimizing feature engineering of a battery BMS management algorithm provided by an embodiment of the present application;

[0051] Figure 2 It is a schematic diagram of a battery BMS management algorithm optimization feature engineering device provided by an embodiment of the present application;

[0052] Figure 3 This is an overall flow chart of a battery BMS management algorithm optimization feature engineering method provided by an embodiment of the present application;

[0053] Figure 4 It is a structural block diagram of a computing device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0054] Many specific details are described in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific implementation disclosed below.

[0055] The terms used in one or more embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit one or more embodiments of the present application. The singular forms of "a", "said" and "the" used in one or more embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in one or more embodiments of the present application refers to and includes any or all possible combinations of one or more associated listed items.

[0056] It should be understood that, although the terms first, second, etc. may be used to describe various information in one or more embodiments of the present application, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0057] In the present application, a method for optimizing feature engineering of a battery BMS management algorithm is provided. The present application also relates to a device for optimizing feature engineering of a battery BMS management algorithm, a computing device, a computer-readable storage medium, and a computer program, which are described in detail one by one in the following embodiments.

[0058] The present invention aims to provide a feature basis for the battery BMS management algorithm. According to the basic knowledge of battery science and artificial intelligence and the experience in algorithm development, a feature engineering device based on the construction and selection of algorithm features of the energy storage battery BMS management system is designed to overcome the difficulty that the original data information is not conducive to directly calculating the battery internal resistance, SOC, SOH and other battery BMS indicators, and provide an efficient and feasible solution. Through the data preprocessing of the battery, the features required for the BMS management related algorithms (internal resistance, SOC, SOH) are constructed to meet the data support for the efficient output of the optimal results of the algorithm development.

[0059] Figure 1 A flowchart of a battery BMS management algorithm optimization feature engineering method provided according to an embodiment of the present application is shown, which specifically includes the following steps.

[0060] Step S101: collecting historical data including cell voltage, cell current and cell temperature of each cell in each battery cluster in the power station, and constructing sample data with equal time intervals based on the historical data;

[0061] Step S102: based on the sample data at equal time intervals, characteristic data including power usage state characteristic data, power usage state segment characteristic data and single power usage cumulative charge and discharge capacity characteristic data are obtained in sequence;

[0062] In one embodiment of the present application, the characteristic data including power usage status characteristic data, power usage status segment characteristic data and single power usage cumulative charge and discharge capacity characteristic data obtained in sequence based on the sample data at equal time intervals include:

[0063] Determine the power consumption status characteristic data of each single battery based on the single battery current in the sample data at equal time intervals;

[0064] According to the power usage state characteristic data of each single battery, the same continuous power usage state characteristic data is divided into a power usage state segment characteristic data, and a power usage state segment number is set for each power usage state segment characteristic data in time order;

[0065] The single-time power consumption cumulative charge-discharge capacity characteristic data of each single battery is obtained by calculating the cumulative charge-discharge capacity of the power consumption status segment characteristic data corresponding to each power consumption status segment number.

[0066] Step S103: based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, construct label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label respectively;

[0067] In one embodiment of the present application, the label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label are constructed based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, respectively, including:

[0068] Based on the charge and discharge SOC in the SOC-OCV table and the voltage value and rated capacity corresponding to each SOC value, convert the charge and discharge SOC into a rated cumulative power consumption capacity, and construct a new SOC-OCV table based on the rated cumulative power consumption capacity;

[0069] Based on the power usage state characteristic data of each single battery and the power usage state segment characteristic data, extracting a first power usage state segment in which the previous power usage state is a static state and the static time length is greater than a preset time length;

[0070] Taking the voltage at the last moment of the previous power usage state as the starting voltage of the first power usage state segment, and determining the rated cumulative power usage capacity corresponding to the starting voltage of the first power usage state segment based on the new SOC-OCV table;

[0071] The rated cumulative power consumption capacity is used as the starting cumulative charging capacity of the first power consumption state segment, and the single cell SOC label is obtained based on the single power consumption cumulative charging and discharging capacity characteristic data of each single cell, the starting cumulative charging capacity of power consumption and the rated capacity data of the single cell.

[0072] In one embodiment of the present application, the label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label are constructed based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, respectively, including:

[0073] Based on the new SOC-OCV table and the single-time power consumption cumulative charge and discharge capacity characteristic data of each single battery, find out the single battery voltage and OCV corresponding to the cumulative power consumption capacity of each power consumption record;

[0074] The internal resistance label of the single cell battery is obtained based on the single cell voltage, OCV and the single cell current corresponding to the cumulative power consumption capacity of each section of the power consumption record.

[0075] In one embodiment of the present application, the label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label are constructed based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, respectively, including:

[0076] Based on the power usage status segmentation feature data, determine each power usage status level, and extract the power usage status segments of the full charging period level and the previous period and the next period being both the deep static period level from all the power usage status levels;

[0077] Obtaining the extracted starting open circuit voltage and ending open circuit voltage of the power usage state segment, and obtaining the starting cumulative charging capacity and ending cumulative charging capacity corresponding to the starting open circuit voltage to the ending open circuit voltage based on the new SOC-OCV table;

[0078] Based on the starting cumulative charging capacity and the ending cumulative charging capacity, the rated cumulative charging capacity change under the rated capacity condition is calculated, and based on the rated cumulative charging capacity change and the cumulative charging capacity, the single cell SOH label is obtained.

[0079] Step S104: performing a BMS management algorithm feature analysis on the historical data, the feature data and the tag data to obtain an optimal feature combination of the BMS management algorithm.

[0080] In one embodiment of the present application, the optimal feature combination of the BMS management algorithm obtained by performing BMS management algorithm feature analysis on the historical data, the feature data and the tag data includes:

[0081] Based on the correlation analysis algorithm, any one of the single cell SOC label, the single cell internal resistance label and the single cell SOH label is used as label data, and any two of them are used as feature data, and the historical data and the feature data are used as feature data, and the correlation coefficient of each feature data and the label data is calculated respectively;

[0082] Based on the random forest algorithm, any one of the single cell SOC label, the single cell internal resistance label and the single cell SOH label is used as label data, and any two of them are used as feature data, and the historical data and the feature data are used as feature data, and the importance index of each feature data and label data is calculated respectively;

[0083] Based on the correlation coefficient and the importance index, the comprehensive correlation index of each feature data and label data is calculated respectively, and the feature data and label data with effective comprehensive correlation index are selected;

[0084] Based on the long short-term memory artificial neural network algorithm and the feature data and label data corresponding to the effective comprehensive correlation index, the optimal feature combination of the BMS management algorithm is obtained.

[0085] In one embodiment of the present application, the feature data and label data corresponding to the long short-term memory artificial neural network algorithm and the effective comprehensive correlation index are used to obtain the optimal feature combination of the BMS management algorithm, including:

[0086] When the label data is a single battery SOC label, based on the long short-term memory artificial neural network algorithm and the selected effective comprehensive correlation index feature data and the single battery SOC label, a feature combination with the smallest SOC model error is calculated, and the feature combination is used as the optimal feature combination of the SOC model;

[0087] When the label data is a single cell internal resistance label, based on the long short-term memory artificial neural network algorithm and the selected effective comprehensive correlation index feature data and the single cell internal resistance label, a feature combination with the minimum error of the internal resistance model is calculated, and the feature combination is used as the optimal feature combination of the internal resistance model;

[0088] When the label data is a single cell SOH label, based on the long short-term memory artificial neural network algorithm and the selected effective comprehensive correlation index feature data and the single cell SOH label, the feature combination with the smallest SOH model error is calculated, and the feature combination is used as the optimal feature combination of the SOH model.

[0089] The preferred feature engineering solution for the battery BMS management algorithm provided in the embodiment of the present application achieves comprehensive analysis of the original data in a specific sequence of steps, provides accurate and efficient data support for the internal resistance algorithm, SOC algorithm, and SOH algorithm required for BMS management, eliminates obstacles to algorithm development caused by data quality issues, saves research and development costs, and avoids interference of poor data quality with the accuracy of the model.

[0090] It should be pointed out that the purpose of this application is to provide high-quality and effective data support for the internal resistance algorithm, SOC algorithm, SOH algorithm, etc. related to the BMS management of energy storage batteries. This method obtains the single cell voltage, total voltage, single cell current, total current, series or parallel relationship of batteries in the battery cluster, SOC-OCV table, single cell rated capacity, and single cell temperature of the battery in the battery cluster in the entire historical period through sampling at the edge of the power station. Based on the knowledge of the battery BMS system, feature engineering and data preprocessing technologies, a battery cluster of a power station is subjected to data preprocessing and feature engineering to produce basic data samples for the battery BMS management algorithm. Among them, all algorithms related to BMS management need to be preprocessed by constructing time series, filtering out abnormal values, and constructing new features to distinguish charging, discharging, and standing, constructing segment number features, respectively constructing charging, discharging, and standing level features, and constructing the cumulative power consumption of each segment. Then, similar redundant features are removed through correlation analysis, random forest algorithm, and long short-term memory artificial neural network (LSTM) algorithm, and finally a feature combination that minimizes the error of the model (internal resistance algorithm, SOC algorithm, SOH algorithm) is obtained. This method provides important data feature support for battery BMS management algorithms such as internal resistance estimation, SOC estimation, and SOH estimation.

[0091] Figure 3 A processing flow chart of a battery BMS management algorithm optimization feature engineering method provided by an embodiment of the present application is shown, which specifically includes the following steps:

[0092] Step 1: Construct time series: The sampling time intervals of the original data are different. In order to meet the algorithm's requirements for equal time intervals of samples, it is necessary to construct sample data with equal time intervals, that is, time series. Create a time data table with 1-minute intervals, splice it with the original data in chronological order, and fill in missing values ​​to integrate the original data into the data table with 1-minute intervals. The output data of this step lays the foundation for the data format of all battery BMS-related algorithms.

[0093] Step 2: Missing value filling: Based on sample data with equal time intervals, use the forward filling method to filter out outliers: those that exceed the normal and reasonable range of battery characteristics, such as the voltage in the original data equal to 0 or greater than 4000.

[0094] Step 3: Construct feature 1: power usage status. Based on the output data of step 2, the power usage status is divided according to the positive and negative signs of the total current (charging: negative current, discharging: positive current, static: current equal to 0).

[0095] Step 4: Construct feature 2: power usage status segmentation. According to the continuity of power usage status in feature 1, the data is segmented. If the data is continuously in the same power usage status, it will be assigned the same segment number. Otherwise, it will be assigned different segment numbers and sorted in chronological order.

[0096] Step 5: Construct feature 3: cumulative charge and discharge capacity of a single power consumption. Based on the power consumption state segmentation of feature 2, the single current calculated by the total current and the series-parallel relationship, and the time interval of the time series in step 1, the cumulative charge and discharge capacity of each power consumption period is calculated through the battery mechanism formula: single current * time interval = charge and discharge capacity of the single battery during this period.

[0097] Step 6: Construct feature 4: Single cell SOC label. Based on the charge and discharge SOC in the SOC-OCV table and the voltage value and rated capacity corresponding to each SOC value, convert the SOC into the rated cumulative power consumption capacity: rated cumulative power consumption capacity = rated capacity * SOC / 100. The SOC-OCV table is usually interval by 5% SOC and the rated cumulative power consumption capacity of 0.1 ampere-hour is used as the sequence interval. The SOC-OCV table is filled by linear interpolation, that is, fitting the rated cumulative power consumption capacity-OCV curve to form a function of OCV = f (rated cumulative power consumption capacity), which is a rated cumulative power consumption capacity of 0.1 ampere-hour for each interval. Calculate the corresponding OCV based on the cumulative power consumption capacity to form a new OCV table. Based on the power consumption status of feature 1 and feature 2, determine whether the previous segment of each segment is static. If so, calculate the static duration. If the static duration is greater than 2 hours, take the voltage at the last moment as the starting voltage of this segment. Based on the new SOC-OCV table, find out the rated cumulative power consumption capacity corresponding to the starting voltage, which is used as the starting cumulative charging capacity of this segment. Based on feature 3, the cumulative charge and discharge capacity of a single power consumption, calculate the SOC label through the ampere-hour integration method: 100% * cumulative charging capacity / rated capacity.

[0098] Step 7: Construct feature 5: power usage status level. Take the starting data record and ending data record of each power usage status in the power usage status segment of feature 2, and according to the single cell SOC label of feature 4, take the starting SOC, ending SOC, starting time point, and ending time point, calculate the SOC change: ending SOC-starting SOC, time change: starting time-ending time, and calculate the power usage status level for each power usage status in the power usage status segment of feature 2. If the value of the power usage status segment of feature 2 is charging or discharging, and the SOC change is greater than the first threshold, then the value of the power usage status level of feature 5 is deep charging or deep discharging, if the SOC change is greater than the second threshold and less than the first threshold, then the value of the power usage status level of feature 5 is shallow charging or shallow discharging, if the SOC change is less than the second threshold and the time consumed is less than the fifth threshold, then the value of the power usage status level of feature 5 is pulse, if the value of the power usage status segment of feature 2 is charging, and the SOC change is greater than the first threshold and the ending SOC is greater than the third threshold, then the power usage status level of feature 5 is The value of the state level is full charge. If the value of the characteristic 2 power state segment is discharge, and the SOC change is greater than the first threshold and the end SOC is less than the fourth threshold, the value of the characteristic 5 power state level is full discharge. If the value of the characteristic 2 power state segment is static, and the time change is greater than the sixth threshold, the value of the characteristic 5 power state level is deep static. If the time change is greater than the seventh threshold and less than the sixth threshold, the value of the characteristic 5 power state level is moderate static. If the time change is less than the seventh threshold, the value of the characteristic 5 power state level is shallow static. The first threshold> the second threshold, the fifth threshold, the third threshold, and the fourth threshold are independent, and the sixth threshold> the seventh threshold.

[0099] Step 8: Construct feature 6: single cell internal resistance label. Based on the new SOC-OCV table output in step 6 and feature 3 single-time cumulative charge and discharge capacity, find the cell voltage and OCV corresponding to the cumulative capacity of each record of each section of electricity consumption, and calculate the internal resistance label through the Rint model: cell voltage-OCV / cell current.

[0100] Step 9: Construct feature 7: single cell SOH label. Based on feature 6 single cell internal resistance label, select the full charge period samples, and then filter out the samples that are deeply static in the previous period and the next period at the same time. According to feature 2 power consumption state segmentation and feature 3 single power consumption cumulative charge and discharge capacity, take the cumulative charging capacity at the end time of each charging period, take the open circuit voltage at the starting point: the end voltage of the previous period, the open circuit voltage at the end point: the end voltage of the next period, based on the new SOC-OCV table output in step 6, find the starting cumulative charging capacity and the ending cumulative charging capacity corresponding to the starting open circuit voltage to the ending open circuit voltage, calculate the rated cumulative charging capacity change under the rated capacity condition, and then calculate the SOH: 100% * cumulative charging capacity / rated cumulative charging capacity change, the SOH of the data segment that does not meet the sample conditions is equal to the SOH of the previous segment.

[0101] Step 10: Feature selection: For SOC, internal resistance, and SOH, feature 4 single cell SOC label, feature 6 single cell internal resistance label, and feature 7 single cell SOH label are used as labels, and other features (single cell voltage, single cell current, single cell temperature, feature 1 power consumption state, feature 2 power consumption state segmentation, feature 3 single power consumption cumulative charge and discharge capacity, feature 4 single cell SOC label / feature 6 single cell internal resistance label / feature 7 single cell SOH label) are used as model features. Based on correlation analysis, the Pearson correlation coefficient of each feature pair label is calculated. Pearson correlation coefficient = covariance of feature and label / (feature standard deviation*label standard deviation), and the correlation coefficients are sorted by absolute value.

[0102] For example, taking feature 7 single cell SOH label as an example, with single cell voltage, single cell current, single cell temperature, feature 1 power usage status, feature 2 power usage status segmentation, feature 3 single power usage cumulative charge and discharge capacity, feature 4 single cell SOC label and feature 6 single cell internal resistance label as features, and feature 7 single cell SOH label as label, based on correlation analysis, calculate the Pearson correlation coefficient of each feature pair label 7 single cell SOH label, Pearson correlation coefficient = covariance of feature and label / (feature standard deviation*label standard deviation), and sort by the absolute value of the correlation coefficient.

[0103] Step 11: Feature selection: Based on the random forest algorithm, features other than labels are used as model features. After preprocessing the training set, test set, and validation set, and data standardization, the random forest model is used for training and the parameters are adjusted to achieve the best model convergence. During the model training process, the importance index of each feature is calculated and sorted by importance index.

[0104] For example, taking feature 7 single cell SOH label as an example, with single cell voltage, single cell current, single cell temperature, feature 1 power state, feature 2 power state segmentation, feature 3 single power consumption cumulative charge and discharge capacity, feature 4 single cell SOC label and feature 6 single cell internal resistance label as features, feature 7 single cell SOH label as label, through training set, test set, validation set segmentation, data standardization preprocessing, using random forest model for training, adjusting parameters to the best model convergence, the importance index of each feature will be calculated during model training, and sorted by importance index;

[0105] Step 12: Comprehensive correlation index = correlation coefficient * 0.5 + importance index * 0.5, select features with correlation index greater than the eighth threshold.

[0106] Step 13: Based on the long short-term memory artificial neural network (LSTM) algorithm, with features other than labels as model features, through preprocessing of training set, test set, validation set segmentation, data standardization, and time sliding window segmentation, traverse the feature combination list to obtain the feature combination that minimizes the model error as the effective feature of the feature single cell SOH algorithm and save it.

[0107] For example, taking feature 7 single cell SOH label as an example, with single cell voltage, single cell current, single cell temperature, feature 1 power usage status, feature 2 power usage status segmentation, feature 3 single power consumption cumulative charging and discharging capacity, feature 4 single cell SOC label, feature 6 single cell internal resistance label as features, feature 7 single cell SOH label as label, through preprocessing of training set, test set, validation set segmentation, data standardization, and time sliding window segmentation, traverse the feature combination list, and obtain the feature combination that minimizes the model error as the effective feature of the feature single cell SOH algorithm, and save it.

[0108] This application also provides an embodiment of a battery BMS management algorithm optimization feature engineering device, Figure 2 The schematic diagram of the structure of a battery BMS management algorithm optimization feature engineering device provided by an embodiment of the present application is shown. Figure 2 As shown, the device comprises:

[0109] The collection module 201 is configured to collect historical data including cell voltage, cell current and cell temperature of each cell in each battery cluster in the power station, and construct sample data with equal time intervals based on the historical data;

[0110] The feature data acquisition module 202 is configured to sequentially obtain feature data including power usage state feature data, power usage state segment feature data and single power usage cumulative charge and discharge capacity feature data based on the sample data at equal time intervals;

[0111] The label data acquisition module 203 is configured to construct label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label respectively based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data;

[0112] The module 204 for obtaining the optimal feature combination is configured to obtain the optimal feature combination of the BMS management algorithm by performing a BMS management algorithm feature analysis on the historical data, the feature data and the tag data.

[0113] The above is a schematic scheme of a battery BMS management algorithm preferred feature engineering device of this embodiment. It should be noted that the technical scheme of the battery BMS management algorithm preferred feature engineering device and the technical scheme of the battery BMS management algorithm preferred feature engineering method mentioned above belong to the same concept. For details not described in detail in the technical scheme of the battery BMS management algorithm preferred feature engineering device, please refer to the description of the technical scheme of the battery BMS management algorithm preferred feature engineering method mentioned above.

[0114] Figure 4 The block diagram of a computing device 400 according to an embodiment of the present application is shown. The components of the computing device 400 include but are not limited to a memory 410 and a processor 420. The processor 420 is connected to the memory 410 via a bus 430, and the database 450 is used to store data.

[0115] The computing device 400 also includes an access device 440 that enables the computing device 400 to communicate via one or more networks 460. Examples of these networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 440 may include one or more of any type of network interface (e.g., a network interface card (NIC)) wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a World Wide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, a near field communication (NFC) interface, and the like.

[0116] In one embodiment of the present application, the above components of the computing device 400 and Figure 4 Other components not shown in the figure may also be connected to each other, for example, via a bus. It should be understood that Figure 4 The computing device structure block diagram shown is only for the purpose of illustration, and is not intended to limit the scope of the present application. Those skilled in the art may add or replace other components as needed.

[0117] Computing device 400 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, a personal digital assistant, a laptop computer, a notebook computer, a netbook, etc.), a mobile phone (e.g., a smart phone), a wearable computing device (e.g., a smart watch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or PC. Computing device 400 may also be a mobile or stationary server.

[0118] The processor 420 is used to execute the following computer executable instructions, which, when executed by the processor, implement the steps of the above-mentioned battery BMS management algorithm optimization feature engineering method.

[0119] The above is a schematic scheme of a computing device of this embodiment. It should be noted that the technical scheme of the computing device and the technical scheme of the above-mentioned battery BMS management algorithm optimization feature engineering method belong to the same concept, and the details not described in detail in the technical scheme of the computing device can be referred to the description of the technical scheme of the above-mentioned battery BMS management algorithm optimization feature engineering method.

[0120] An embodiment of the present application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the above-mentioned battery BMS management algorithm optimization feature engineering method.

[0121] The above is a schematic scheme of a computer-readable storage medium of this embodiment. It should be noted that the technical scheme of the storage medium and the technical scheme of the above-mentioned battery BMS management algorithm preferred feature engineering method belong to the same concept, and the details not described in detail in the technical scheme of the storage medium can be referred to the description of the technical scheme of the above-mentioned battery BMS management algorithm preferred feature engineering method.

[0122] An embodiment of the present application further provides a computer program, wherein when the computer program is executed in a computer, the computer is caused to execute the steps of the above-mentioned battery BMS management algorithm optimization feature engineering method.

[0123] The above is a schematic scheme of a computer program of this embodiment. It should be noted that the technical scheme of the computer program and the technical scheme of the battery BMS management algorithm optimization feature engineering method mentioned above belong to the same concept, and the details not described in detail in the technical scheme of the computer program can be found in the description of the technical scheme of the battery BMS management algorithm optimization feature engineering method mentioned above.

[0124] The above describes specific embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do 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.

[0125] The computer instructions include computer program codes, which may be in source code form, object code form, executable files or some intermediate forms, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0126] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present application are not limited by the described action sequence, because according to the embodiments of the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of the present application.

[0127] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0128] The preferred embodiments of the present application disclosed above are only used to help explain the present application. The optional embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of the embodiments of the present application. The present application selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of the present application, so that those skilled in the art can understand and use the present application well. The present application is only limited by the claims and their full scope and equivalents.

Claims

1. A method for optimizing feature engineering of a battery BMS management algorithm, characterized in that: include: Collect historical data including cell voltage, cell current and cell temperature of each cell in each battery cluster in the power station, and construct sample data with equal time intervals based on the historical data; Based on the sample data at equal time intervals, characteristic data including power usage status characteristic data, power usage status segment characteristic data and single power usage cumulative charge and discharge capacity characteristic data are obtained in sequence; Based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label are constructed respectively; By performing a BMS management algorithm feature analysis on the historical data, the feature data and the tag data, an optimal feature combination of the BMS management algorithm is obtained.

2. The method according to claim 1, characterized in that The characteristic data including power consumption state characteristic data, power consumption state segment characteristic data and single power consumption cumulative charge and discharge capacity characteristic data obtained in sequence based on the sample data of equal time intervals include: Determine the power consumption status characteristic data of each single battery based on the single battery current in the sample data at equal time intervals; According to the power usage state characteristic data of each single battery, the same continuous power usage state characteristic data is divided into a power usage state segment characteristic data, and a power usage state segment number is set for each power usage state segment characteristic data in time order; The single-time power consumption cumulative charge-discharge capacity characteristic data of each single battery is obtained by calculating the cumulative charge-discharge capacity of the power consumption status segment characteristic data corresponding to each power consumption status segment number.

3. The method according to claim 1 or 2, characterized in that: The label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label are constructed based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, respectively, including: Based on the charge and discharge SOC in the SOC-OCV table and the voltage value and rated capacity corresponding to each SOC value, convert the charge and discharge SOC into a rated cumulative power consumption capacity, and construct a new SOC-OCV table based on the rated cumulative power consumption capacity; Based on the power usage state characteristic data of each single battery and the power usage state segment characteristic data, extracting a first power usage state segment in which the previous power usage state is a static state and the static time length is greater than a preset time length; Taking the voltage at the last moment of the previous power usage state as the starting voltage of the first power usage state segment, and determining the rated cumulative power usage capacity corresponding to the starting voltage of the first power usage state segment based on the new SOC-OCV table; The rated cumulative power consumption capacity is used as the starting cumulative charging capacity of the first power consumption state segment, and the single cell SOC label is obtained based on the single power consumption cumulative charging and discharging capacity characteristic data of each single cell, the starting cumulative charging capacity of power consumption and the rated capacity data of the single cell.

4. The method according to claim 3, characterized in that The label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label are constructed based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, respectively, including: Based on the new SOC-OCV table and the single-time power consumption cumulative charge and discharge capacity characteristic data of each single battery, find out the single battery voltage and OCV corresponding to the cumulative power consumption capacity of each power consumption record; The internal resistance label of the single cell battery is obtained based on the single cell voltage, OCV and the single cell current corresponding to the cumulative power consumption capacity of each section of the power consumption record.

5. The method according to claim 4, characterized in that The label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label are constructed based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data, respectively, including: Based on the power usage status segmentation feature data, determine each power usage status level, and extract the power usage status segments of the full charging period level and the previous period and the next period being both the deep static period level from all the power usage status levels; Obtaining the extracted starting open circuit voltage and ending open circuit voltage of the power usage state segment, and obtaining the starting cumulative charging capacity and ending cumulative charging capacity corresponding to the starting open circuit voltage to the ending open circuit voltage based on the new SOC-OCV table; Based on the starting cumulative charging capacity and the ending cumulative charging capacity, the rated cumulative charging capacity change under the rated capacity condition is calculated, and based on the rated cumulative charging capacity change and the cumulative charging capacity, the single cell SOH label is obtained.

6. The method according to claim 5, characterized in that The optimal feature combination of the BMS management algorithm obtained by performing a BMS management algorithm feature analysis on the historical data, the feature data and the tag data includes: Based on the correlation analysis algorithm, any one of the single cell SOC label, the single cell internal resistance label and the single cell SOH label is used as label data, and any two of them are used as feature data, and the historical data and the feature data are used as feature data, and the correlation coefficient of each feature data and the label data is calculated respectively; Based on the random forest algorithm, any one of the single cell SOC label, the single cell internal resistance label and the single cell SOH label is used as label data, and any two of them are used as feature data, and the historical data and the feature data are used as feature data, and the importance index of each feature data and label data is calculated respectively; Based on the correlation coefficient and the importance index, the comprehensive correlation index of each feature data and label data is calculated respectively, and the feature data and label data with effective comprehensive correlation index are selected; Based on the long short-term memory artificial neural network algorithm and the feature data and label data corresponding to the effective comprehensive correlation index, the optimal feature combination of the BMS management algorithm is obtained.

7. The method according to claim 6, characterized in that The optimal feature combination of the BMS management algorithm obtained by the feature data and label data corresponding to the long short-term memory artificial neural network algorithm and the effective comprehensive correlation index includes: When the label data is a single battery SOC label, based on the long short-term memory artificial neural network algorithm and the selected effective comprehensive correlation index feature data and the single battery SOC label, a feature combination with the smallest SOC model error is calculated, and the feature combination is used as the optimal feature combination of the SOC model; When the label data is a single cell internal resistance label, based on the long short-term memory artificial neural network algorithm and the selected effective comprehensive correlation index feature data and the single cell internal resistance label, a feature combination with the minimum error of the internal resistance model is calculated, and the feature combination is used as the optimal feature combination of the internal resistance model; When the label data is a single cell SOH label, based on the long short-term memory artificial neural network algorithm and the selected effective comprehensive correlation index feature data and the single cell SOH label, the feature combination with the smallest SOH model error is calculated, and the feature combination is used as the optimal feature combination of the SOH model.

8. A battery BMS management algorithm optimization feature engineering device, characterized in that: include: A collection module is configured to collect historical data including cell voltage, cell current and cell temperature of each cell in each battery cluster in the power station, and construct sample data with equal time intervals based on the historical data; The characteristic data acquisition module is configured to sequentially obtain characteristic data including power consumption state characteristic data, power consumption state segment characteristic data and single power consumption cumulative charge and discharge capacity characteristic data based on the sample data at equal time intervals; A label data acquisition module is configured to construct label data including a single cell SOC label, a single cell internal resistance label and a single cell SOH label respectively based on the power consumption state characteristic data, the power consumption state segment characteristic data and the single power consumption cumulative charge and discharge capacity characteristic data in the characteristic data; The module for obtaining the optimal feature combination is configured to obtain the optimal feature combination of the BMS management algorithm by performing a BMS management algorithm feature analysis on the historical data, the feature data and the tag data.

9. A computing device comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the battery BMS management algorithm optimization feature engineering method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the battery BMS management algorithm optimization feature engineering method described in any one of claims 1 to 7.

Images