CN-122017627-A - Battery fault monitoring method, system, equipment and medium based on voltage characteristics

CN122017627ACN 122017627 ACN122017627 ACN 122017627ACN-122017627-A

Abstract

The invention provides a battery fault monitoring method, a system, equipment and a medium based on voltage characteristics, wherein the monitoring method comprises the steps of extracting battery data of a plurality of continuous charging/discharging periods from battery historical data; according to the voltage data of each cell, the longitudinal outlier average value and the average voltage ranking of each cell in each charging/discharging period are calculated, a plurality of charging/discharging periods are screened out from all charging/discharging periods according to the average voltage ranking, the longitudinal outlier average value of all charging/discharging periods is screened out to form a longitudinal outlier average value sequence, the average voltage ranking of all charging/discharging periods is screened out to form an average voltage ranking sequence, and the longitudinal outlier average value sequence and the average voltage ranking sequence of each cell are input into a pre-trained monitoring model to generate a monitoring result of each cell in a battery. According to the invention, whether the battery has faults or not is analyzed, so that the safety of the battery in the use process is improved.

Inventors

YANG HENGZHAO
ZHAO JIAQI

Assignees

上海科技大学

Dates

Publication Date: 20260512
Application Date: 20260224

Claims (10)

1. A battery fault monitoring method based on voltage characteristics, comprising: Extracting battery data of a plurality of continuous charging/discharging periods from the battery history data, wherein the battery data comprises voltage data of each electric core in the battery in the charging/discharging periods; According to the voltage data of each cell, calculating the longitudinal outlier average value and the average voltage ranking of each cell in each charging/discharging period; Screening a plurality of charging/discharging time periods from all charging/discharging time periods according to the average voltage ranking aiming at each battery cell, forming a longitudinal outlier average value sequence by using the longitudinal outlier average values of all the screened charging/discharging time periods, and forming an average voltage ranking sequence by using the average voltage ranking of all the screened charging/discharging time periods; and inputting the longitudinal outlier average value sequence and the average voltage ranking sequence of each cell into a pre-trained monitoring model to generate a monitoring result of each cell in the battery.
2. The method for monitoring battery faults based on voltage characteristics of claim 1, wherein the voltage data comprises voltages of the respective cells at respective sampling points, and the calculating a longitudinal outlier average of each cell in respective charge/discharge periods according to the voltage data of each cell comprises: Calculating longitudinal outliers of the sampling points of each battery cell in each charging/discharging period; According to the longitudinal outlier and the sampling point number of all the sampling points of each battery cell in each charging/discharging period, calculating the longitudinal outlier average value of each battery cell in each charging/discharging period; Longitudinal outliers The method comprises the following steps: ; longitudinal outlier mean The method comprises the following steps: ; Wherein, the Represented as the first in the battery The electric core is provided with a plurality of electric cores, Denoted as the first In the case of a single charge/discharge period, Denoted as the first The first charge/discharge period A number of sampling points are used to sample the sample, Represented as at the first The first sampling point The voltage of the electric core is controlled by the voltage of the electric core, Represented as at the first The voltage intermediate value of all the battery cells at each sampling point, Denoted as the first Number of sampling points in each charge/discharge period.
3. The method for monitoring battery faults based on voltage characteristics of claim 1, wherein said voltage data comprises voltages of respective cells at respective sampling points, said calculating an average voltage ranking of each cell over respective charge/discharge periods from the voltage data of each cell comprises: ordering the voltages of all the battery cells at each sampling point in each charging/discharging period, and recording the voltage ranking of each sampling point of each battery cell in each charging/discharging period; calculating the average voltage ranking of each battery cell in each charging/discharging period according to the voltage ranking and the sampling point number of all sampling points of each battery cell in each charging/discharging period; Average voltage ranking The method comprises the following steps: ; Wherein, the Represented as the first in the battery The electric core is provided with a plurality of electric cores, Denoted as the first In the case of a single charge/discharge period, Denoted as the first The first charge/discharge period A number of sampling points are used to sample the sample, Represented as at the first The first sampling point The voltage ranking of the individual cells is determined, Denoted as the first Number of sampling points in each charge/discharge period.
4. The method for monitoring battery faults based on voltage characteristics according to claim 1, wherein for each cell, screening a plurality of charging/discharging periods from all charging/discharging periods according to an average voltage ranking, forming a longitudinal outlier average sequence from longitudinal outlier averages of all charging/discharging periods screened, forming an average voltage ranking sequence from average voltage ranks of all charging/discharging periods screened, comprising: calculating a difference value between the average voltage rank of each charging/discharging period and the average voltage rank of the previous charging/discharging period for each cell, and recording the charging/discharging period with the negative difference value and the largest absolute value as an intermediate charging/discharging period; Sequentially selecting a preset number of charging/discharging periods from all the charging/discharging periods according to the positions of the intermediate charging/discharging periods; The vertical outlier average of all charging/discharging periods is selected to form a vertical outlier average sequence, and the average voltage of all charging/discharging periods is selected to form an average voltage ranking sequence.
5. The method for monitoring battery faults based on voltage characteristics of claim 4, wherein selected charge/discharge periods satisfy: ; Wherein, the Indicated as the starting position of the selected charge/discharge period in all charge/discharge periods; indicated as the position of the intermediate charge/discharge period in all charge/discharge periods; Indicated as selecting a corresponding preset number of charge/discharge periods.
6. The battery fault monitoring method based on voltage characteristics according to claim 1, wherein the monitoring model is trained by the following method: acquiring a training sample set, wherein the training sample set comprises a longitudinal outlier average value sequence and an average voltage ranking sequence of a plurality of electric cores and sample labels of the electric cores; inputting the longitudinal outlier average value sequences and the average voltage ranking sequences of the plurality of battery cells in the training sample set into a monitoring model to be trained to obtain a corresponding prediction result, wherein the monitoring model is a one-dimensional convolutional neural network model; Calculating a loss value of the monitoring model loss function according to sample labels and prediction results of the plurality of battery cells; and based on the loss value, carrying out parameter adjustment on the monitoring model to be trained to obtain the monitoring model after training.
7. The voltage signature based battery fault monitoring method of claim 6 wherein the loss function is expressed as: ; Wherein, the Represented as a training sample set The average loss value of the individual cells, Represented as the first of the training sample set The electric core is provided with a plurality of electric cores, A weight factor that is represented as a fault sample configuration, Denoted as the first The sample labels of the individual cells are then read, Represented as a logarithmic function; is expressed as monitoring model prediction No The probability that each cell is a fault cell is 0-1; A weight factor that is represented as a normal sample configuration.
8. A battery fault monitoring system based on voltage characteristics, comprising: An extracting unit for extracting battery data of a plurality of continuous charge/discharge periods from the battery history data, the battery data including voltage data of each cell in the battery during the charge/discharge periods; the computing unit is used for computing the longitudinal outlier average value and the average voltage ranking of each battery cell in each charging/discharging period according to the voltage data of each battery cell; The sequence forming unit is used for screening a plurality of charging/discharging time periods from all charging/discharging time periods according to the average voltage ranking aiming at each battery cell, forming a longitudinal outlier average value sequence by the longitudinal outlier average value of all the screened charging/discharging time periods, and forming an average voltage ranking sequence by the average voltage ranking of all the screened charging/discharging time periods; And the generating unit is used for inputting the longitudinal outlier average value sequence and the average voltage ranking sequence of each battery core into a pre-trained monitoring model to generate a monitoring result of each battery core in the battery.
9. An electronic device, the electronic device comprising: one or more processors; Storage means for storing one or more programs that, when executed by the one or more processors, cause the electronic device to implement the voltage feature-based battery fault monitoring method of any of claims 1-7.
10. A computer-readable storage medium, having stored thereon a computer program which, when executed by a processor of a computer, causes the computer to perform the battery fault monitoring method based on voltage characteristics of any one of claims 1 to 7.

Description

Battery fault monitoring method, system, equipment and medium based on voltage characteristics Technical Field The present invention relates to the field of battery technologies, and in particular, to a method, a system, an apparatus, and a medium for monitoring battery faults based on voltage characteristics. Background The battery is used as an electric energy storage component, comprises, but is not limited to, a lead-acid battery, a lithium ion battery (such as a lithium iron phosphate battery, a ternary lithium battery, a lithium titanate battery and the like), a sodium ion battery, a lithium metal battery, a semi-solid battery, a solid-state battery and the like, and is widely applied to various electric scenes. Such as electrified vehicles, e.g., electric vehicles, electric ships, electric airplanes, and energy storage systems, e.g., energy storage power stations, data centers, intelligent computing centers. In practical application, internal short circuit fault of a battery is one of main causes of thermal runaway of the battery, is usually caused by growth of lithium dendrite or defects of a diaphragm, has extremely weak initial characteristics, and is extremely easy to evolve into serious internal short circuit to cause fire accident. Since the initial characteristics of the short-circuit fault in the battery are extremely weak, it is difficult to determine the short-circuit fault directly from the parameters of the battery. Therefore, there is a need for improvement. Disclosure of Invention The invention provides a battery fault monitoring method, a system, equipment and a medium based on voltage characteristics, which are used for solving the technical problems that the initial characteristics of the short circuit faults in the existing battery are very weak and are difficult to judge directly through the parameters of the battery. The invention provides a battery fault monitoring method based on voltage characteristics, which comprises the following steps: Extracting battery data of a plurality of continuous charging/discharging periods from the battery history data, wherein the battery data comprises voltage data of each electric core in the battery in the charging/discharging periods; According to the voltage data of each cell, calculating the longitudinal outlier average value and the average voltage ranking of each cell in each charging/discharging period; Screening a plurality of charging/discharging time periods from all charging/discharging time periods according to the average voltage ranking aiming at each battery cell, forming a longitudinal outlier average value sequence by using the longitudinal outlier average values of all the screened charging/discharging time periods, and forming an average voltage ranking sequence by using the average voltage ranking of all the screened charging/discharging time periods; and inputting the longitudinal outlier average value sequence and the average voltage ranking sequence of each cell into a pre-trained monitoring model to generate a monitoring result of each cell in the battery. In one embodiment of the invention, the voltage data comprises voltages of the battery cells at the sampling points, and the calculating the longitudinal outlier average value of each battery cell in the charging/discharging time period according to the voltage data of each battery cell comprises the following steps: Calculating longitudinal outliers of the sampling points of each battery cell in each charging/discharging period; According to the longitudinal outlier and the sampling point number of all the sampling points of each battery cell in each charging/discharging period, calculating the longitudinal outlier average value of each battery cell in each charging/discharging period; Longitudinal outliers The method comprises the following steps: ; longitudinal outlier mean The method comprises the following steps: ; Wherein, the Represented as the first in the batteryThe electric core is provided with a plurality of electric cores,Denoted as the firstIn the case of a single charge/discharge period,Denoted as the firstThe first charge/discharge periodA number of sampling points are used to sample the sample,Represented as at the firstThe first sampling pointThe voltage of the electric core is controlled by the voltage of the electric core,Represented as at the firstThe voltage intermediate value of all the battery cells at each sampling point,Denoted as the firstNumber of sampling points in each charge/discharge period. In one embodiment of the invention, the voltage data comprises voltages of the battery cells at the sampling points, and the calculating the average voltage rank of each battery cell in the charging/discharging period according to the voltage data of each battery cell comprises the following steps: ordering the voltages of all the battery cells at each sampling point in each charging/discharging period, and recording the voltage ranking of each samp