Battery inconsistency diagnosis method based on capacity estimation

Abstract

The invention discloses a battery inconsistency diagnosis method based on capacity estimation, and the method comprises the steps: obtaining the historical data of a vehicle, and carrying out the preprocessing of the historical data; carrying out correlation analysis on the preprocessed data, screening out data fields of which the correlation with the capacity of the battery monomer exceeds a set correlation coefficient, and marking the data fields as health factors; constructing a battery cell capacity prediction model based on a time sequence prediction algorithm, and training the battery cell capacity prediction model by using the screened field data and historical capacity data; acquiring real-time data of the vehicle, picking out screened data fields, transmitting the data into the trained battery cell capacity prediction model, and predicting to obtain the capacity of the battery cell; and carrying out abnormal coefficient calculation on the predicted capacity of the single battery, and if the abnormal coefficient is not in a preset range, judging that the battery module has an inconsistency fault. According to the invention, whether the power battery has an inconsistency fault or not can be accurately judged.

Description

Technical Field

[0001] The present invention relates to the field of power battery safety, and in particular to a battery inconsistency diagnosis method based on capacity estimation. Background Art

[0002] Electric vehicles have been widely recognized as an important part of the efficient and sustainable transportation system of the future. The battery system is a key part of electric vehicles, which largely determines the driving performance and the mileage per charge. The battery system is structurally composed of a series of battery cells connected in different series and parallel modes, the purpose of which is to provide the required voltage and capacity respectively. Despite the continuous advancement of battery manufacturing and packaging technology, each component battery or related accessories may suffer from loss due to the aging process or abuse during operation. Undetected faults can have a harmful effect on the battery and even lead to catastrophic accidents under extreme conditions. The inconsistency of each cell in the battery module is one of the common battery failures, including voltage inconsistency, temperature inconsistency and capacity inconsistency. Early and accurate detection of battery inconsistency failures helps to take proactive measures to avoid serious damage. Therefore, it is very important and meaningful to diagnose battery inconsistency failures in a timely and accurate manner in actual operation.

[0003] Most of the existing battery fault diagnosis methods are based on voltage and temperature. They do not analyze the battery inconsistency problem from the perspective of battery charge and discharge capacity. The analysis angle is relatively single and difficult to apply in practice. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention proposes a battery inconsistency diagnosis method based on capacity estimation. The method predicts the capacity of the battery cell in the next charge and discharge cycle through a machine learning method, and at the same time determines whether the battery module has an inconsistency fault through the abnormal coefficient of the capacity, thereby realizing battery module inconsistency fault diagnosis.

[0005] The purpose of the present invention is achieved through the following technical solutions:

[0006] A battery inconsistency diagnosis method based on capacity estimation, the method comprising the following steps:

[0007] S1: Obtain historical data of the vehicle and pre-process it to obtain as much and comprehensive valid data as possible;

[0008] S2: Perform correlation analysis on the preprocessed data, screen out data fields whose correlation with the battery cell capacity exceeds a set correlation coefficient, and mark them as health factors;

[0009] S3: Build a battery cell capacity prediction model based on the time series prediction algorithm, and use the field data filtered by S2 and the historical capacity data to train the battery cell capacity prediction model;

[0010] S4: Acquire the real-time data of the vehicle, select the data field screened in step S2, and pass it into the trained battery cell capacity prediction model to predict the capacity of the battery cell in the next charge and discharge cycle;

[0011] S5: Calculate the abnormal coefficient of the capacity of the battery cell in the next charge and discharge cycle predicted by S4. If the abnormal coefficient is not within a preset range, it is determined that the battery module has an inconsistency fault.

[0012] Furthermore, the historical data of the vehicle in S1 includes average charge and discharge voltage, total charging time, voltage peak, zero crossing frequency of battery impedance phase, temperature, constant current charging time, instantaneous voltage drop at the beginning of discharge, and open circuit voltage.

[0013] Furthermore, the correlation analysis in S2 is selected from any one of person correlation coefficient, spearman correlation coefficient and kendall correlation coefficient.

[0014] Furthermore, the value range of the correlation coefficient set in S2 is 0.5-0.6.

[0015] Furthermore, the time series prediction algorithm in S3 is selected from LSTM algorithm or Informer algorithm.

[0016] Furthermore, the calculation formula of the capacity abnormality coefficient of each battery cell in S5 is:

[0017]

[0018] Among them, C refers to the capacity of the battery cell, C ave Refers to the average capacity of all battery cells, σ C Refers to the standard deviation of the capacity of all battery cells.

[0019] Furthermore, in S5, the preset range of the capacity abnormality coefficient is obtained by statistically analyzing historical data, calculating the capacity abnormality data of historical normal modules and inconsistent modules, and statistically analyzing to obtain the threshold value of the capacity abnormality coefficient.

[0020] Furthermore, in S1, the method of preprocessing the historical data of the vehicle includes data cleaning and / or data analysis.

[0021] The beneficial effects of the present invention are as follows:

[0022] The present invention makes full use of the correlation function to screen the battery health factor, uses big data, i.e. the historical data of the power battery, to train the time series prediction model, and uses the model to predict the capacity of the battery cell in the next cycle. At the same time, the battery capacity is used to diagnose the battery inconsistency fault. The method of the present invention can accurately determine whether the battery has an inconsistency fault accurately and efficiently, and avoid subsequent serious damage. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A flow chart of a battery inconsistency diagnosis method based on capacity estimation provided by an embodiment of the present invention.

[0024] Figure 2 This is a flow chart of obtaining a battery capacity abnormal coefficient threshold value based on historical data in an embodiment. DETAILED DESCRIPTION

[0025] The present invention will be described in detail below based on the accompanying drawings and preferred embodiments, and the purpose and effects of the present invention will become more clear. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0026] like Figure 1 As shown, the battery inconsistency diagnosis method based on capacity estimation provided by an embodiment of the present invention includes the following steps:

[0027] Step 1: Obtain historical data of single cells in the battery module and preprocess it.

[0028] In this embodiment, the historical data of the vehicle is obtained from the data platform, and the historical data of the vehicle includes but is not limited to: average charge and discharge voltage, total charging time, voltage peak, zero crossing frequency of battery impedance phase, temperature, constant current charging time, instantaneous voltage drop at the beginning of discharge, open circuit voltage, etc. The specific historical data and real-time data of the vehicle are associated with the driving data of the vehicle and are selected according to the actual situation. The preprocessing method includes data cleaning and / or data analysis, so as to obtain as much and comprehensive valid data as possible.

[0029] Step 2: Perform correlation analysis on the preprocessed data, filter out data fields whose correlation with the battery cell capacity exceeds the set correlation coefficient, and mark them as health factors. The correlation analysis can use any of the person correlation coefficient, spearman correlation coefficient, and kendall correlation coefficient.

[0030] In this embodiment, the person correlation coefficient analysis method is used to screen out data fields that have a certain correlation with the battery capacity and mark them as health factors, and the health factors are used as inputs to the time series prediction algorithm. The correlation coefficient value range set in this embodiment is 0.5-0.6.

[0031] Step 3: Build a battery cell capacity prediction model based on the time series prediction algorithm, and use the health factor and historical capacity data to train the prediction model. The time series prediction algorithm here includes the LSTM (Long sequence time-series forecasting) algorithm or the Informer algorithm.

[0032] Step 4: Obtain the real-time data of the vehicle, select the data fields filtered out in step 2, pass the data into the trained battery cell capacity prediction model, and predict the capacity of the battery cell in the next charge and discharge cycle;.

[0033] Step 5: Calculate the abnormal coefficient of the predicted battery cell capacity. If the abnormal coefficient is not within the preset range, it is determined that the battery module is inconsistent.

[0034] The calculation formula of the abnormal coefficient of each battery cell capacity is:

[0035]

[0036] Among them, C refers to the capacity of the battery cell, C ave Refers to the average value of all single battery capacities, σ C Refers to the standard deviation of the capacity of all single cells.

[0037] Figure 2 A flowchart for obtaining the battery capacity abnormal coefficient threshold value based on historical data is provided. The battery capacity abnormal coefficient in the historical data is calculated; the normal value range of the battery capacity abnormal coefficient A of the single cell voltage is obtained by statistically analyzing the battery normal data and the inconsistency fault data. When the abnormal coefficient calculated in real time based on the input single cell capacity exceeds the corresponding normal value range, it is determined that the battery has an inconsistency fault.

[0038] In this example, the value range of the battery capacity abnormal coefficient A is [-4, 3], that is, the value range of A exceeds [-4, 3], and it is determined that an inconsistency fault exists in the power battery.

[0039] It should be noted that the above numerical values ​​-4 and 3 are for illustration only and are not intended to limit the protection scope of the present invention.

[0040] The present invention makes full use of the correlation function to screen the battery health factor, uses big data, i.e. the historical data of the power battery, to train the time series prediction model, and uses the model to predict the charge and discharge capacity of the battery cell in the next cycle. At the same time, the battery capacity is used to diagnose the battery inconsistency fault. The method of the present invention can accurately determine whether there is an inconsistency fault in the battery, which is accurate and efficient, avoids subsequent serious damage, and has strong foresight.

[0041] Those skilled in the art can understand that the above are only preferred examples of the invention and are not intended to limit the invention. Although the invention is described in detail with reference to the above examples, those skilled in the art can still modify the technical solutions recorded in the above examples or replace some of the technical features therein with equivalents. Any modification, equivalent replacement, etc. made within the spirit and principle of the invention shall be included in the protection scope of the invention.

Claims

1. A battery inconsistency diagnosis method based on capacity estimation, characterized in that: The method comprises the following steps: S1: Obtain historical data of the vehicle and pre-process it to obtain as much and comprehensive valid data as possible; S2: Perform correlation analysis on the preprocessed data, screen out data fields whose correlation with the battery cell capacity exceeds a set correlation coefficient, and mark them as health factors; S3: Build a battery cell capacity prediction model based on the time series prediction algorithm, and use the field data filtered by S2 and the historical capacity data to train the battery cell capacity prediction model; S4: Acquire the real-time data of the vehicle, select the data field screened in step S2, and pass it into the trained battery cell capacity prediction model to predict the capacity of the battery cell in the next charge and discharge cycle; S5: Calculate the abnormal coefficient of the capacity of the battery cell in the next charge and discharge cycle predicted by S4. If the abnormal coefficient is not within a preset range, it is determined that the battery module has an inconsistency fault.

2. The battery inconsistency diagnosis method based on capacity estimation according to claim 1, characterized in that: The historical data of the vehicle in S1 includes average charge and discharge voltage, total charging time, voltage peak, zero-crossing frequency of battery impedance phase, temperature, constant current charging time, instantaneous voltage drop at the beginning of discharge, and open circuit voltage.

3. The battery inconsistency diagnosis method based on capacity estimation according to claim 1, characterized in that: The correlation analysis in S2 is selected from any one of the person correlation coefficient, the spearman correlation coefficient and the kendall correlation coefficient.

4. The battery inconsistency diagnosis method based on capacity estimation according to claim 1, characterized in that: The value range of the correlation coefficient set in S2 is 0.5-0.

6.

5. The battery inconsistency diagnosis method based on capacity estimation according to claim 1, characterized in that: The time series prediction algorithm in S3 is selected from LSTM algorithm or Informer algorithm.

6. The battery inconsistency diagnosis method based on capacity estimation according to claim 1, characterized in that: The calculation formula of the capacity abnormality coefficient of each battery cell in S5 is: Among them, C refers to the capacity of the battery cell, C ave Refers to the average capacity of all battery cells, σ C Refers to the standard deviation of the capacity of all battery cells.

7. The battery inconsistency diagnosis method based on capacity estimation according to claim 4, characterized in that: In S5, the preset range of the capacity abnormality coefficient is obtained by statistically analyzing historical data, calculating the capacity abnormality data of historical normal modules and inconsistent modules, and obtaining the threshold value of the capacity abnormality coefficient through statistical analysis.

8. The battery inconsistency diagnosis method based on capacity estimation according to claim 1, characterized in that: In S1, the method of preprocessing the historical data of the vehicle includes data cleaning and / or data analysis.

Images