A method, system, device and storage medium for evaluating the health of an energy storage battery

Abstract

The present invention discloses a method for evaluating the health state of an energy storage battery, which specifically includes data acquisition, data preprocessing, correction of the estimated maximum battery capacity, correction of the calibrated battery capacity value, calculation of the corrected battery SOH value, feature mining, construction of a short-term SOH prediction model, and construction of a long-term SOH prediction model. The method and system for evaluating the health state of the energy storage battery collect a large amount of historical charge and discharge data of the battery. Based on the historical charge and discharge data of a large number of energy storage batteries, the estimation of the maximum battery capacity and the calibrated battery capacity value are corrected, so that the accuracy of the battery SOH value can be further improved, and the historical data for training the SOH prediction model is also more accurate, thereby further improving the prediction accuracy of the short-term SOH prediction model and the long-term SOH prediction model. The method for evaluating the health state of the energy storage battery can simultaneously implement two methods of short-term SOH evaluation and long-term SOH evaluation, and can achieve a better evaluation effect on the battery SOH.

Description

Technical Field

[0001] This patent relates to the technical field of energy storage battery management systems, and particularly to a method for detecting and evaluating the state of health (SOH) of batteries. Background Art

[0002] With the increasing shortage of energy and the continuous improvement of people's environmental awareness, various new energy technologies have received more and more attention, and the new energy field has entered a peak period of development. Energy storage is an important part of the new energy field, and the research on energy storage batteries as energy storage carriers has also been continuously deepened. Among them, the state of health (SOH) of the battery is an important indicator of the battery and an important reference for whether the battery can operate safely. SOH estimation is a method used to judge the health status of the battery, expressed as the percentage of the charge or discharge capacity of the battery to the nominal capacity of the battery. For a new battery, this value is generally greater than 100%. As the battery is used and ages, this value gradually decreases. According to the IEEE standard 1188-1996, when the capacity of the power battery drops to 80%, it should be replaced. Therefore, the accuracy of the energy storage battery health assessment method plays a crucial role in the safe use of the battery.

[0003] Traditional methods for evaluating the health of lithium batteries usually first collect historical data of the battery SOH (state of health of the battery), then use the ampere-hour integration method to calculate the charging power of a certain charging segment, and then divide it by the difference in the SOC (current remaining charge of the battery) change interval corresponding to this charging segment to obtain the maximum capacity of the battery. Then, the maximum capacity of the battery is divided by the rated capacity of the battery to obtain the battery SOH. This method for detecting and evaluating the state of health (SOH) of the battery has problems such as low prediction accuracy and poor real-time performance. With the continuous development of battery charging technology, this method for detecting and evaluating the state of health of the battery can no longer meet the usage requirements. Therefore, it is necessary to design an energy storage battery health assessment method that can better achieve accurate prediction of the state of health of the battery. Summary of the Invention

[0004] To overcome the deficiencies in the above-mentioned prior art, the purpose of the present invention is to design a method for more accurately evaluating the health of energy storage batteries.

[0005] An energy storage battery health assessment method includes the following steps:

[0006] 1) Data collection: Collect and record the historical charge and discharge data of the battery at regular time intervals. The historical charge and discharge data of the battery includes battery charging current, battery discharge current, battery charging voltage, battery discharge voltage, battery charging temperature, battery discharge temperature, and battery SOC parameters;

[0007] 2) Data preprocessing: Perform data deduplication, null value processing, and abnormal value processing on the acquired historical charge and discharge data of the battery;

[0008] 3) Maximum battery capacity estimation correction: According to the historical charge and discharge data of the battery, divide the battery SOC from 0% to 100% into multiple charging intervals at equal proportional intervals, calculate the average value of the actual charging power in each charging interval, and then divide it by the average value of the charging power when the battery is fully charged to obtain the corresponding table of valuation conversion factors for the charging intervals. Multiply the actual charging amount in each charging interval covered by the charging by the corresponding valuation conversion factor of that charging interval and sum them up. Then divide the sum value by the sum of the valuation conversion factors of the covered charging intervals to obtain the corrected estimated value of the maximum battery capacity;

[0009] 4) Battery calibrated capacity value correction: Divide the calibrated capacity corresponding to each charging rate and temperature marked at the time of battery factory by the battery correction capacity coefficient L to obtain the capacity conversion factor corresponding to each charging rate and temperature marked at the time of battery factory. Use the equal division interpolation method to calculate the capacity conversion factors at each interpolation point between the capacity conversion factors corresponding to two adjacent temperatures at each charging rate, and form a capacity conversion factor table. Multiply the capacity conversion factor corresponding to the current ambient temperature and charging rate by the battery correction capacity coefficient L to obtain the corrected battery calibrated capacity;

[0010] 5) Calculate the corrected battery SOH value: Divide the corrected estimated value of the maximum battery capacity by the corrected battery calibrated capacity;

[0011] 6) Feature mining: Mine the historical charge and discharge data of the battery to obtain useful features. The useful features include calendar days, cumulative usage duration, number of charging times, number of cycles, number of deep charge and discharge times, average temperature, average temperature difference, maximum temperature difference, charging unit temperature rise amount, standard deviation of average temperature difference, average pressure difference, maximum pressure difference, standard deviation of average pressure difference, large charging current records, and large discharge current records;

[0012] 7) Build a short-term SOH prediction model: Build a short-term SOH prediction model based on the corrected battery SOH value and the mined useful features and train it;

[0013] 8) Build a long-term SOH prediction model: Build a long-term SOH prediction model with SOH time series data, and use the prediction results of the short-term SOH prediction model as training samples to train the long-term SOH prediction model.

[0014] Preferably, in step 7), build two prediction models, the LightGBM model and the CatBoost model, to predict the short-term SOH value respectively, and take the average value of the prediction results of the two models to obtain the final short-term SOH value.

[0015] Preferably, the long-term SOH prediction model constructed in step 8) is an LSTM prediction model.

[0016] Preferably, the number of neurons in the LSTM prediction model is 16; the optimizer uses RMSprop, the learning rate is 0.0001, and the loss function is mae.

[0017] To achieve the above object, the present invention also discloses a system for evaluating the health of an energy storage battery, a data acquisition unit, the data acquisition unit is used to collect and record the historical charge and discharge data of the battery at a certain time interval, and the historical charge and discharge data of the battery includes battery charging current, battery discharge current, battery charging voltage, battery discharge voltage, battery charging temperature, battery discharge temperature, battery SOC parameter; a data preprocessing unit, the data preprocessing unit is used to perform data deduplication, null value processing and abnormal value processing on the acquired historical charge and discharge data of the battery; a maximum battery capacity estimation correction unit, the maximum battery capacity estimation correction unit is used to divide the battery SOC from 0% to 100% into multiple charging intervals at equal proportional intervals according to the historical charge and discharge data of the battery, calculate the average value of the actual charging power of each charging interval, and then divide it by the average value of the charging power when the battery is fully charged, to obtain the corresponding table of the estimation conversion factor of the charging interval, multiply the actual charging amount of each charging interval covered by the charge by the corresponding estimation conversion factor of the charging interval and sum them, and then divide the value after summation by the sum of the estimation conversion factors of the covered charging intervals, to obtain the corrected estimated value of the maximum battery capacity; a battery calibrated capacity value correction unit, the battery calibrated capacity value correction unit is used to divide the calibrated capacity corresponding to each charge rate and temperature marked when the battery leaves the factory by a battery correction capacity coefficient L to obtain the capacity conversion factor corresponding to each charge rate and temperature marked when the battery leaves the factory, and use the equal division interpolation method between the capacity conversion factors corresponding to two adjacent temperatures at each charge rate to calculate the capacity conversion factors at each interpolation point, and form a capacity conversion factor table, multiply the capacity conversion factor corresponding to the current ambient temperature and charge rate by the battery correction capacity coefficient L to obtain the corrected battery calibrated capacity; a battery SOH value correction calculation unit, the battery SOH value correction calculation unit is used to divide the corrected maximum battery capacity estimation value by the corrected battery calibrated capacity to obtain the battery SOH value correction value; a feature mining unit, the feature mining unit is used to mine the historical charge and discharge data of the battery to obtain useful features, and the useful features include calendar days, cumulative usage duration, number of charges, number of cycles, number of deep charge and discharge times, average temperature, average temperature difference, maximum temperature difference, charging unit temperature rise amount, average temperature difference standard deviation, average pressure difference, maximum pressure difference, average pressure difference standard deviation, charging large current record and discharge large current record; a short-term SOH prediction model, the short-term SOH prediction model is constructed based on the corrected battery SOH value and the mined useful features, and is used to predict the short-term SOH of the battery; a long-term SOH prediction model, the long-term SOH prediction model is constructed with SOH time series data, and is trained with the prediction results of the short-term SOH prediction model as test samples, and is used to predict the long-term SOH.

[0018] To achieve the above object, the present invention also discloses a device for evaluating the health of an energy storage battery, which includes: a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

[0019] To achieve the above object, the present invention also discloses a computer storage medium for storing a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

[0020] The above technical solution has the following beneficial effects: The method and system for evaluating the health of an energy storage battery collect a large amount of historical charge and discharge data of the battery. Based on the historical charge and discharge data of a large number of energy storage batteries, a valuation conversion factor table for each charging segment of the energy storage battery is constructed to correct the estimated maximum battery capacity and improve the accuracy of the maximum battery capacity valuation; by constructing capacity conversion factors for different temperatures and charging rates of the energy storage battery, the calibrated capacity of the battery is also more accurate, so that the accuracy of the corrected battery SOH value can be further improved, and the historical data for training the SOH prediction model is also more accurate, thereby further improving the prediction accuracy of the short-term SOH prediction model and the long-term SOH prediction model. The method for evaluating the health of an energy storage battery can simultaneously implement two methods of short-term SOH evaluation and long-term SOH evaluation, and can achieve a better evaluation effect on the battery SOH. Description of the Drawings

[0021] Figure 1 It is a flowchart of the evaluation method according to an embodiment of the present invention.

[0022] Figure 2 It is a sample of the corresponding table of the valuation conversion factor for the SOC charging interval.

[0023] Figure 3 It is a sample of the calibrated capacity marked when the battery leaves the factory.

[0024] Figure 4 It is a sample of the battery capacity conversion factor table.

[0025] Figure 5 It is a short-term SOH prediction model diagram of the battery according to an embodiment of the present invention.

[0026] Figure 6 It is a long-term SOH prediction model diagram of the battery according to an embodiment of the present invention.

[0027] Figure 7 Sample of predicting short-term SOH by LightGBM model.

[0028] Figure 8 Sample of predicting short-term SOH by CatBoost model.

[0029] Figure 9 Variation diagram of the loss function for predicting the long-term SOH of the LSTM model.

[0030] Figure 10 Trend diagram of the long-term SOH predicted by the LSTM model. Specific implementation mode

[0031] The following specific embodiments illustrate the implementation mode of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification.

[0032] Now, the exemplary embodiments will be described more fully with reference to the accompanying drawings. However, the exemplary embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the present invention will be more complete and comprehensive, and the concept of the exemplary embodiments will be fully conveyed to those skilled in the art. The features, structures, or characteristics described can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of the embodiments of the present invention. However, those skilled in the art will realize that the technical solutions of the present invention can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, well-known technical solutions are not shown or described in detail to avoid obscuring the various aspects of the present invention.

[0033] As Figure 1 shown, the present invention discloses a method for evaluating the health state of an energy storage battery, which specifically includes the following steps:

[0034] 1. Data collection:

[0035] Collect the historical charge and discharge data of the battery. The historical charge and discharge data of the battery include the battery charging current, battery discharge current, battery charging voltage, battery discharge voltage, battery charging temperature, battery discharge temperature, and battery SOC parameter. When collecting data, it is collected at a certain time interval and stored for recording.

[0036] 2. Data preprocessing:

[0037] Clean the collected historical charge and discharge data of the battery. Because after the original data is collected, there will inevitably be some dirty data, such as missing data, abnormal data, etc. It is necessary to clean the original data and perform data deduplication, null value processing, and abnormal value processing on the obtained historical charge and discharge data of the battery.

[0038] 3. Maximum battery capacity estimation correction:

[0039] According to the historical charge and discharge data of the battery, the battery SOC is divided into multiple charging intervals at equal proportional intervals from 0% to 100%. For example, each interval of 10% is used as an interval, that is, the SOC distribution of the interval is 0%-10%, 10%-20%,..., 90%-100%. Then, the valuation conversion factor of each interval is calculated. The specific calculation method is to calculate the average value of the actual charging power of each interval based on a large amount of battery historical charging data, and then divide it by the average value of the charging power when the battery is fully charged, so as to obtain the valuation conversion factor of this interval. An example of the SOC interval conversion factor table is as Figure 2 shown. When calculating the actual maximum capacity of the battery, assume that the SOC change range during the battery charging process covers several SOC intervals. Then, multiply the actual charging power of each SOC interval by the valuation conversion factor corresponding to this SOC interval, and then sum them up. Then, divide the sum value by the sum of the valuation conversion factors of the covered SOC intervals, so as to obtain the final estimated value of the battery maximum capacity.

[0040] 4. Battery calibration capacity value correction:

[0041] Divide the calibration capacity corresponding to each charge rate and temperature marked at the time of battery factory by the battery correction capacity coefficient L to obtain the capacity conversion factor corresponding to each charge rate and temperature marked at the time of battery factory. Use the equal division interpolation method to calculate the capacity conversion factors at each interpolation point between the capacity conversion factors corresponding to two adjacent temperatures at each charge rate, and form a capacity conversion factor table. Multiply the capacity conversion factor corresponding to the current ambient temperature and charge rate by the battery correction capacity coefficient L to obtain the corrected battery calibration capacity. In order to understand the specific correction method of this step more clearly, the following example further illustrates this correction method.

[0042] Such as Figure 3As shown, it is a sample of the calibrated capacity of a storage battery when it leaves the factory. The charging rates covered by the calibrated capacity marked on this storage battery are 0.2C, 0.5C, 1C, 2C, and 5C, and the temperatures are generally -20°, 0°, 20°, and 45°. At the same charge-discharge rate, the calibrated capacity at 25 degrees is the highest, and those at 0 degrees and 45 degrees are both on the low side. Taking 25 degrees as the center point, it gradually decreases towards both ends. Because in the actual charging process of the battery, the temperature range is relatively wide and is not limited to the above several temperature values, it is necessary to expand the temperature value range to better conform to the temperature conditions during actual charging. The value of the battery correction capacity coefficient L can be set according to the capacity of the battery. Generally, an integer value close to the battery capacity can be selected. In this embodiment, the battery correction capacity coefficient L can be set to 38. Divide the calibrated capacity corresponding to each charging rate and temperature by the battery correction capacity coefficient L to obtain the capacity conversion factor corresponding to each charging rate and temperature marked on the battery when it leaves the factory. Then, use the equal division interpolation method between the capacity conversion factors corresponding to two adjacent temperatures at each charging rate to calculate the capacity conversion factors at each interpolation point and form a capacity conversion factor table. The equal division quantity of the equal division interpolation method can be set according to the required temperature accuracy for detection.

[0043] As Figure 4 shown, in the present invention, the temperature interval is refined to 1 degree, that is, the interval formed by the capacity conversion factors corresponding to 20° and 0° respectively is evenly divided into 20 equal parts by inserting equal division points, and the capacity conversion factor values corresponding to each equal division point are calculated. In the same way, the capacity conversion factor values at different charging rates are calculated, and finally, a capacity conversion factor table as Figure 4 shown can be obtained. Then, by multiplying the capacity conversion factor corresponding to the previous ambient temperature and charging rate by the battery correction capacity coefficient L, the corrected current battery calibrated capacity can be obtained.

[0044] 5. Calculate the corrected battery SOH value:

[0045] Dividing the estimated value of the corrected maximum battery capacity by the corrected battery calibrated capacity can obtain the corrected battery SOH value.

[0046] 6. Feature mining:

[0047] The historical data of the battery charge-discharge data records the original features such as current, voltage, and temperature. Through corresponding calculations on the historical charge-discharge data of the battery, useful features for constructing the SOH prediction model are mined. The useful features specifically include: calendar days, cumulative usage duration, number of charging times, number of cycles, number of deep charge-discharge times, average temperature, average temperature difference, maximum temperature difference, temperature rise amount per unit of charge, standard deviation of average temperature difference, average pressure difference, maximum pressure difference, standard deviation of average pressure difference, large charging current record, and large discharging current record.

[0048] 7. Build a short-term SOH prediction model:

[0049] Build a short-term SOH prediction model based on the corrected battery SOH value and the mined useful features. Use the useful features as the input and the corrected battery SOH value as the output to train the model. As Figure 5 shown, in order to further improve the prediction accuracy of the model, the present invention constructs two LightGBM models and CatBoost models to respectively predict the short-term SOH value, and takes the average of the prediction results of the two models to obtain the final short-term SOH value.

[0050] Both LightGBM and CatBoost are one of the gradient boosting algorithms and are improved versions of the GBDT (Gradient Boosting Decision Tree) algorithm. The advantage of LightGBM is that it supports high-efficiency parallel training, has a faster training speed, lower memory consumption, better accuracy, and supports distributed processing of massive data. The main advantages of the CatBoost algorithm are the processing of categorical features, and there is no need to process categorical features through feature engineering before training the model; another advantage is the processing of prediction offsets, which reduces overfitting of the model and improves the model prediction effect.

[0051] 8. Build a long-term SOH prediction model:

[0052] As Figure 6 shown, construct a long-term SOH prediction model with the battery historical SOH value as time series data, and use the prediction result of the short-term SOH prediction model as the training sample to train the long-term SOH prediction model. In this way, the long-term SOH of the battery can be predicted through this long-term SOH prediction model. As a specific implementation manner, the long-term SOH prediction model adopts an LSTM prediction model. LSTM (Long Short-Term Memory) is a long short-term memory neural network model, which is a type of time recurrent neural network that improves the long-term dependence problem existing in the recurrent neural network (RNN). The number of neurons in the adopted LSTM model is 16; the optimizer adopts RMSprop, the learning rate is 0.0001, and the loss function is mae.

[0053] To verify the prediction effect of the above model, this patent conducts a model effect test based on the actual energy storage data of a certain company. The test is divided into two parts, one is short-term SOH prediction, and the other is long-term SOH prediction.

[0054] Short-term SOH prediction effect: Train the LightGBM and CatBoost models respectively, and use the mean absolute error as the model evaluation index. The calculation results are as follows:

[0055] Table 1 Display of Model Effect Indicators

[0056] Model Mean Absolute Error LightGBM 0.013 CatBoost 0.009

[0057] The prediction results of the LightGBM model are as follows Figure 7 shown; the prediction results of the CatBoost model are as Figure 8 shown; it can be seen from the evaluation indicators and prediction examples that the prediction effect of short-term SOH is relatively ideal.

[0058] Long-term SOH prediction effect: The change of the model loss function is as follows Figure 9 shown, and the training loss curve and validation loss curve of this model are basically the same. The display of the model prediction effect is as Figure 10 shown, where the error between the true historical SOH value and the historical SOH value predicted by the model is very small and the stability is good. It can be seen from the figure that as time goes by, the predicted future SOH value shows a downward trend, which also conforms to the actual situation of the battery. The prediction effect of this model for long-term SOH is also relatively ideal.

[0059] The present invention also discloses a system for evaluating the health of an energy storage battery, which includes: a data acquisition unit for collecting and recording the historical charge and discharge data of the battery at a certain time interval. The historical charge and discharge data of the battery includes battery charging current, battery discharge current, battery charging voltage, battery discharge voltage, battery charging temperature, battery discharge temperature, and battery SOC parameter;

[0060] A data preprocessing unit for performing data deduplication, null value processing, and abnormal value processing on the obtained historical charge and discharge data of the battery;

[0061] A maximum battery capacity estimation correction unit for dividing the battery SOC from 0% to 100% into multiple charging intervals at equal proportional intervals according to the historical charge and discharge data of the battery, calculating the average value of the actual charging power of each charging interval, and then dividing it by the average value of the charging power when the battery is fully charged to obtain the corresponding table of valuation conversion factors for the charging intervals. Multiply the actual charging amount of each charging interval covered by the charge by the valuation conversion factor corresponding to this charging interval and sum them up, and then divide the value after summation by the sum of the valuation conversion factors of the covered charging intervals to obtain the corrected estimated value of the maximum battery capacity;

[0062] The battery calibrated capacity value correction unit is used to divide the calibrated capacity corresponding to each charge rate and temperature of the battery by the calibrated capacity at the time of battery factory shipment. Then, within each temperature range of each charge rate, the equal division interpolation method is adopted to obtain the conversion factor for each temperature of each charge rate, and a conversion factor table of the battery's temperature and charge-discharge rate with respect to the calibrated capacity is obtained. Multiply the estimated value of the battery calibrated capacity by the conversion factor of the corresponding temperature and charge-discharge rate with respect to the calibrated capacity to obtain the corrected battery calibrated capacity; the battery SOH value correction calculation unit is used to divide the corrected maximum battery capacity estimate by the corrected battery calibrated capacity to obtain the battery SOH value correction value;

[0063] The feature mining unit is used to mine the historical charge-discharge data of the battery to obtain useful features. The useful features include calendar days, cumulative usage duration, number of charge times, number of cycles, number of deep charge-discharge times, average temperature, average temperature difference, maximum temperature difference, temperature rise per charging unit, standard deviation of average temperature difference, average pressure difference, maximum pressure difference, standard deviation of average pressure difference, large charging current records, and large discharging current records;

[0064] The short-term SOH prediction model is constructed based on the corrected battery SOH value and the mined useful features, and is used to predict the short-term SOH of the battery;

[0065] The long-term SOH prediction model is constructed with SOH time series data and trained with the prediction results of the short-term SOH prediction model as test samples, and is used to predict the long-term SOH.

[0066] It should be noted that for other corresponding descriptions of each functional unit involved in the energy storage battery health assessment system provided in this embodiment, reference can be made to the corresponding descriptions of the energy storage battery health assessment method, which will not be elaborated here.

[0067] Based on the above energy storage battery health assessment method, to achieve the above object, an embodiment of the present application also provides an energy storage battery health assessment device, which can specifically be a personal computer, a server, a network device, etc. This physical device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to implement the above energy storage battery health assessment method.

[0068] Correspondingly, this embodiment also provides a computer storage medium, on which a computer program is stored. When the program is executed by a processor, the above-mentioned energy storage battery health assessment method is implemented. Based on such an understanding, the technical solution of this application can be embodied in the form of a software product. This software product can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various implementation scenarios of this application. The storage medium may also include an operating system and a network communication module. The operating system is a program for managing the hardware and software resources of the above-mentioned computer device, and supports the operation of information processing programs and other software and / or programs. The network communication module is used to implement communication between components inside the storage medium, as well as communication between this entity device and other hardware and software.

[0069] The energy storage battery health assessment method constructs an SOH prediction model by using a machine learning model, uses a large number of original features and mined features, and the models used are the LightGBM and CatBoost algorithms. The prediction results of the two are averaged to make the overall effect optimal. Based on the charge and discharge historical data of a large number of energy storage batteries, a conversion factor table for each charging segment is constructed, and the estimated maximum battery capacity is corrected by the ampere-hour integration method to improve the accuracy of the battery maximum capacity estimation. By constructing different temperatures and charging rates, the calibrated capacity of the battery is also made more accurate. After improving the accuracy of both the battery maximum capacity estimation and the calibrated capacity, the historical data used for the SOH prediction model training is also more accurate.

[0070] The energy storage battery health assessment method and system collect a large amount of historical charge and discharge data of the battery. Based on the charge and discharge historical data of a large number of energy storage batteries, the battery maximum capacity estimation and the battery calibrated capacity value are corrected, which can further improve the accuracy of the battery SOH value and make the historical data for the SOH prediction model training more accurate, thereby further improving the prediction accuracy of the short-term SOH prediction model and the long-term SOH prediction model. The energy storage battery health assessment method can simultaneously implement two methods of short-term SOH assessment and long-term SOH assessment, and can achieve a better assessment effect on the battery SOH.

[0071] The above embodiments are only illustrative of the principles and effects of the present invention, and are not used to limit the present invention. Any person familiar with this technology can modify or change the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or changes completed by those with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed by the present invention should still be covered by the claims of the present invention.

Claims

1. A method for evaluating the health state of an energy storage battery, characterized in that, It includes the following steps: 1) Data collection: Collect and record the historical charge and discharge data of the battery at regular time intervals. The historical charge and discharge data of the battery includes battery charging current, battery discharge current, battery charging voltage, battery discharge voltage, battery charging temperature, battery discharge temperature, and battery SOC parameters; 2) Data preprocessing: Perform data deduplication, null value processing, and abnormal value processing on the obtained historical charge and discharge data of the battery; 3) Maximum battery capacity estimation correction: According to the historical charge and discharge data of the battery, divide the battery SOC from 0% to 100% into multiple charging intervals at equal proportional intervals, calculate the average value of the actual charging power of each charging interval, and then divide it by the average value of the charging power when the battery is fully charged to obtain the corresponding table of the estimation conversion factor for the charging interval. Multiply the actual charging amount of each charging interval covered by the charge by the corresponding estimation conversion factor of this charging interval and sum them up. Then divide the sum value by the sum of the estimation conversion factors of the covered charging intervals to obtain the corrected estimated value of the maximum battery capacity; 4) Battery calibrated capacity value correction: Divide the calibrated capacity corresponding to each charge rate and temperature marked at the time of battery factory by the battery correction capacity coefficient L to obtain the capacity conversion factor corresponding to each charge rate and temperature marked at the time of battery factory. Use the equal division interpolation method to calculate the capacity conversion factors at each interpolation point between the capacity conversion factors corresponding to two adjacent temperatures at each charge rate and form a capacity conversion factor table. Multiply the capacity conversion factor corresponding to the current ambient temperature and charge rate by the battery correction capacity coefficient L to obtain the corrected battery calibrated capacity; The battery correction capacity coefficient L is selected as an integer value close to the battery capacity; 5) Calculate the corrected SOH value of the battery: Divide the corrected estimated value of the maximum battery capacity by the corrected battery calibrated capacity; 6) Feature mining: Mine the historical charge and discharge data of the battery to obtain useful features. The useful features include calendar days, cumulative usage duration, number of charges, number of cycles, number of deep charge and discharge times, average temperature, average temperature difference, maximum temperature difference, charging unit temperature rise, standard deviation of average temperature difference, average pressure difference, maximum pressure difference, standard deviation of average pressure difference, large charging current record, and large discharge current record; 7) Build a short-term SOH prediction model: Build and train a short-term SOH prediction model based on the corrected SOH value of the battery and the mined useful features; 8) Build a long-term SOH prediction model: Build a long-term SOH prediction model with SOH time series data, and use the prediction results of the short-term SOH prediction model as training samples to train the long-term SOH prediction model.

2. The method for evaluating the state of health of an energy storage battery according to claim 1, wherein, wherein, In step 7), two prediction models, the LightGBM model and the CatBoost model, are built to predict the short-term SOH value respectively, and the average value of the prediction results of the two models is calculated to obtain the final short-term SOH value.

3. The method for evaluating the state of health of an energy storage battery according to claim 1, wherein The long-term SOH prediction model built in step 8) is an LSTM prediction model.

4. The method for evaluating the state of health of an energy storage battery according to claim 3, wherein The number of neurons in the LSTM prediction model is 16; the optimizer uses RMSprop, the learning rate is 0.0001, and the loss function is mae.

5. A state of health assessment system for an energy storage battery, characterized in that, It includes: A data acquisition unit, which is used to collect and record the historical charge and discharge data of the battery at certain time intervals. The historical charge and discharge data of the battery includes battery charging current, battery discharging current, battery charging voltage, battery discharging voltage, battery charging temperature, battery discharging temperature, and battery SOC parameter; A data preprocessing unit, which is used to perform data deduplication, null value processing, and abnormal value processing on the obtained historical charge and discharge data of the battery; A maximum battery capacity estimation correction unit, which is used to divide the battery SOC from 0% to 100% into multiple charging intervals at equal proportional intervals according to the historical charge and discharge data of the battery, calculate the average value of the actual charging power of each charging interval, then divide it by the average value of the charging power when the battery is fully charged, obtain the corresponding table of the estimation conversion factor for the charging interval, multiply the actual charging amount of each charging interval covered by the charge by the corresponding estimation conversion factor of this charging interval and sum them up, and then divide the value after summation by the sum of the estimation conversion factors of the covered charging intervals to obtain the corrected estimated value of the maximum battery capacity; A battery calibrated capacity value correction unit, which is used to divide the calibrated capacity corresponding to each charge rate and temperature marked at the time of battery factory by the battery correction capacity coefficient L to obtain the capacity conversion factor corresponding to each charge rate and temperature marked at the time of battery factory. The equal division interpolation method is used to calculate the capacity conversion factors at each interpolation point between the capacity conversion factors corresponding to two adjacent temperatures at each charge rate, and form a capacity conversion factor table. Multiply the capacity conversion factor corresponding to the current ambient temperature and charge rate by the battery correction capacity coefficient L to obtain the corrected battery calibrated capacity; The battery correction capacity coefficient L selects an integer value close to the battery capacity; A battery SOH value correction calculation unit, which is used to divide the corrected maximum battery capacity estimation value by the corrected battery calibrated capacity to obtain the battery SOH value correction value; A feature mining unit, which is used to mine the historical charge and discharge data of the battery to obtain useful features. The useful features include calendar days, cumulative usage duration, number of charges, number of cycles, number of deep charge and discharge times, average temperature, average temperature difference, maximum temperature difference, charging unit temperature rise amount, average temperature difference standard deviation, average pressure difference, maximum pressure difference, average pressure difference standard deviation, charging large current record, and discharging large current record; A short-term SOH prediction model, which is constructed based on the corrected battery SOH value and the mined useful features and is used to predict the short-term SOH of the battery; A long-term SOH prediction model, which is constructed with SOH time series data and trained with the prediction results of the short-term SOH prediction model as training samples and is used to predict the long-term SOH.

6. An equipment for evaluating the health state of an energy storage battery, which comprises a memory and a processor, and a computer program is stored in the memory, characterized in that, When the processor executes the computer program, it realizes the steps of the method described in any one of claims 1 to 4.

7. A computer storage medium for storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4.

Images