CN-122016129-A - Battery pack pressure distribution dynamic monitoring method based on multi-sensor data fusion

Abstract

The invention relates to the technical field of data processing, in particular to a battery pack pressure distribution dynamic monitoring method based on multi-sensor data fusion, which comprises the steps of obtaining a self-adaptive working condition window of each moment according to the stability of the variation trend of working condition data of each working condition index of a target battery pack in a local time range of each moment; based on the data similarity of various working condition indexes in the self-adaptive working condition window of each historical moment and the current moment, the operation working condition similarity of each historical moment and the current moment is obtained, and then the pressure at each monitoring point of the target battery pack at each historical moment is combined, the pressure standard value at each monitoring point of the target battery pack is obtained, and according to the difference between the pressure at each monitoring point of the target battery pack at the current moment and the pressure standard value, the abnormal pressure distribution of the target battery pack at the current moment is monitored, so that the accuracy of real-time monitoring of the abnormal pressure distribution of the battery pack is improved.

Inventors

• CHEN YUSEN

Assignees

• 宁德三化智能科技有限公司

Dates

Publication Date

20260512

Application Date

20260416

Claims (8)

1. 1. The battery pack pressure distribution dynamic monitoring method based on the multi-sensor data fusion is characterized by comprising the following steps of: acquiring at least two working condition indexes for reflecting different operation working conditions of a target battery pack, and working condition data of each working condition index at each historical moment in a preset time period at and before the current moment; Aiming at any historical moment, according to the change trend stability of the working condition data of each working condition index in the local time range of any historical moment, acquiring an adaptive working condition window of any historical moment, wherein a target battery pack under each moment in the adaptive working condition window is judged to be in the same operating condition; Acquiring a self-adaptive working condition window of each historical moment and the current moment, and acquiring the operation working condition similarity of each historical moment and the current moment based on the data similarity of each working condition index in the self-adaptive working condition window of each historical moment and the current moment; according to the similarity of the operation working conditions of each historical moment and the current moment and the pressure of each monitoring point of the target battery pack at each historical moment, acquiring a pressure standard value of each monitoring point of the target battery pack, acquiring the pressure of each monitoring point of the target battery pack at the current moment, and according to the difference between the pressure of each monitoring point of the target battery pack at the current moment and the pressure standard value, monitoring the abnormal pressure distribution of the target battery pack at the current moment in real time.
2. 2. The method for dynamically monitoring the pressure distribution of the battery pack based on the multi-sensor data fusion according to claim 1, wherein the obtaining the adaptive working condition window at any historical moment according to the stability of the variation trend of the working condition data of each working condition index in the local time range at any historical moment comprises: For any working condition index, forming an initial window of the any historical moment and two moments closest to the any historical moment, performing linear fitting on working condition data of the any working condition index at each moment in the initial window to obtain fitting data of the any working condition index at each moment in the initial window, and acquiring the fitting goodness of the any working condition index in the initial window according to the difference between the working condition data of the any working condition index at each moment in the initial window and the fitting data, wherein the fitting goodness is recorded as initial fitting goodness; Respectively acquiring the time adjacent to the starting time and the ending time of the initial window, marking the time as the adjacent time of the initial window, adding any adjacent time into the initial window to obtain a new window of any historical time, and performing linear fitting on the working condition data of any working condition index at each time in the new window to obtain the fitting goodness of any working condition index in the new window, and marking the fitting goodness as a new fitting goodness; subtracting the initial fitting goodness from the new fitting goodness to obtain a target difference, and carrying out normalization processing on the ratio between the target difference and the initial fitting goodness to obtain the data trend consistency degree of the any working condition index between any adjacent moment and each moment in the initial window; acquiring a new window corresponding to another adjacent moment, and acquiring the data trend consistency degree of any working condition index between the other adjacent moment and each moment in the initial window according to the fitting goodness of any working condition index in the new window corresponding to the other adjacent moment; if the data trend consistency degree corresponding to any adjacent moment is greater than the preset data trend consistency degree threshold value, taking a new window corresponding to the adjacent moment corresponding to the maximum data trend consistency degree as an initial window of any historical moment, if only the data trend consistency degree corresponding to one adjacent moment is greater than the preset data trend consistency degree threshold value, taking the new window corresponding to the adjacent moment of which the data trend consistency degree is greater than the preset data trend consistency degree threshold value as the initial window of any historical moment, and repeating the step of acquiring the data trend consistency degree corresponding to the adjacent moment of the initial window until the data trend consistency degree corresponding to each adjacent moment of the initial window is less than or equal to the preset data trend consistency degree threshold value, and taking the initial window as the working condition window of any working condition index at any historical moment; the working condition windows of all working condition indexes at any historical moment are obtained, windows corresponding to intersections of all working condition windows are recorded as public windows, if the time length corresponding to the public windows is greater than or equal to a preset minimum analysis time length, the public windows are used as self-adaptive working condition windows at any historical moment, and if the time length corresponding to the public windows is less than the preset minimum analysis time length, the left and right time sequence directions of the public windows are uniformly expanded to the preset minimum analysis time length, so that the self-adaptive working condition windows at any historical moment are obtained.
3. 3. The method for dynamically monitoring the pressure distribution of the battery pack based on the multi-sensor data fusion according to claim 2, wherein the step of obtaining the goodness of fit of any one of the operating conditions in the initial window according to the difference between the operating condition data and the fitting data of the operating conditions in each time in the initial window comprises the following steps: Calculating the absolute value of the difference value between the working condition data and the fitting data of any working condition index at each moment in the initial window, and taking the opposite number of the average value of all the absolute values of the difference value as the independent variable of the natural exponential function to obtain the fitting goodness of any working condition index in the initial window.
4. 4. The method for dynamically monitoring the pressure distribution of the battery pack based on the multi-sensor data fusion according to claim 1, wherein the step of obtaining the operation condition similarity between each historical moment and the current moment based on the data similarity between each working condition index in the adaptive working condition window between each historical moment and the current moment comprises the following steps: Clustering the working condition data of each working condition index at the current moment and each historical moment respectively to obtain at least two clusters corresponding to each working condition index, and obtaining the reflecting weights of each working condition index on different operation working conditions of the target battery pack according to the data stability and the inter-cluster gaps in the clusters corresponding to each working condition index; For any historical moment, working condition data of any working condition index at each moment in an adaptive working condition window of any historical moment are formed into a data sequence, the data sequence is recorded as a historical data sequence, the data sequence of any working condition index in the adaptive working condition window of the current moment is obtained, the data sequence is recorded as a current data sequence, the DTW distance between the historical data sequence and the current data sequence is calculated, and the reciprocal between the DTW distance and a preset constant is used as the data similarity of any working condition index between the any historical moment and the current moment; The data similarity of each working condition index between any historical moment and the current moment is obtained, the data similarity of each working condition index between any historical moment and the current moment is weighted and summed according to the reflection weight of each working condition index, and the weighted and summed result is normalized to obtain the operation working condition similarity of any historical moment and the current moment.
5. 5. The method for dynamically monitoring the pressure distribution of the battery pack based on the multi-sensor data fusion according to claim 4, wherein the step of obtaining the reflecting weights of the various operating conditions of the target battery pack according to the data stability and the inter-cluster gaps in the clusters corresponding to the various operating conditions comprises the following steps: Aiming at any working condition index, acquiring a clustering center of each cluster corresponding to the any working condition index, calculating standard deviations of all clustering centers to obtain inter-cluster standard deviations, and respectively acquiring standard deviations of data in each cluster corresponding to the any working condition index to obtain intra-cluster standard deviations of each cluster corresponding to the any working condition index; Obtaining maximum values from all intra-cluster standard deviations and the inter-cluster standard deviations, marking the maximum standard deviations as the maximum standard deviations, and carrying out normalization processing on the duty ratio of the inter-cluster standard deviations in the maximum standard deviations to obtain first distinguishing factors; The ratio of the standard deviation in the cluster of each cluster corresponding to any working condition index in the maximum standard deviation is calculated respectively, and the opposite number of the mean value of the ratio corresponding to all clusters is used as an independent variable of a natural exponential function to obtain a second distinguishing factor; Calculating the product between the first distinguishing factor and the second distinguishing factor to obtain the distinguishing degree of any working condition index on different operation working conditions of a target battery pack; the distinguishing degree of each working condition index on different operation working conditions of the target battery pack is obtained, the duty ratio of the distinguishing degree corresponding to any working condition index in the accumulation sum of all distinguishing degrees is calculated, and the reflecting weight of any working condition index on different operation working conditions of the target battery pack is obtained.
6. 6. The method for dynamically monitoring the pressure distribution of the battery pack based on the multi-sensor data fusion according to claim 1, wherein the step of obtaining the pressure standard value at each monitoring point of the target battery pack according to the similarity of the operation condition of each historical time and the current time and the pressure at each monitoring point of the target battery pack at each historical time comprises the following steps: the pressure at each monitoring point of the target battery pack at each historical moment is respectively obtained, the standard deviation of the pressure at all monitoring points of the target battery pack at each historical moment is respectively calculated, and the method is utilized Acquiring normal standard deviation ranges of all pressure standard deviations in principle, and eliminating historical moments when the pressure standard deviation exceeds the normal standard deviation ranges to obtain target historical moments; And aiming at any monitoring point of the target battery pack, taking the similarity of the operation working conditions of each target historical moment and the current moment as weight, calculating the weighted average value of the pressure at any monitoring point under all target historical moments, and recording the weighted average value as the pressure standard value at any monitoring point of the target battery pack.
7. 7. The method for dynamically monitoring the pressure distribution of the battery pack based on the multi-sensor data fusion according to claim 1, wherein the monitoring the abnormality of the pressure distribution of the target battery pack at the current moment in real time according to the difference between the pressure at each monitoring point of the target battery pack at the current moment and the pressure standard value comprises the following steps: For any monitoring point of a target battery pack, acquiring real-time pressure at the any monitoring point at the current time, calculating a difference absolute value between a pressure standard value at the any monitoring point and the real-time pressure, and taking the duty ratio of the difference absolute value in the pressure standard value at the any monitoring point as the pressure abnormality degree at the any monitoring point at the current time; And acquiring the pressure abnormality degree of any monitoring point at each historical time, acquiring an abnormality upper limit from the pressure abnormality degree of any monitoring point at all the historical time by using a box diagram, and if the pressure abnormality degree of any monitoring point at the current time is greater than the abnormality upper limit, carrying out pressure abnormality alarm on any monitoring point of the target battery pack.
8. 8. The method for dynamically monitoring the pressure distribution of a battery pack based on multi-sensor data fusion according to claim 1, wherein the operating condition index comprises the current, the state of charge of the target battery pack and the temperature at each monitoring point of the target battery pack.

Description

Battery pack pressure distribution dynamic monitoring method based on multi-sensor data fusion Technical Field The invention relates to the technical field of data processing, in particular to a battery pack pressure distribution dynamic monitoring method based on multi-sensor data fusion. Background With the wide application of power batteries in the fields of new energy automobiles, energy storage systems and the like, the safety and stability of battery packs are increasingly concerned in the operation monitoring of related systems. When the battery pack is in faults such as overcharging, overheating and internal short circuit, the problems of abnormal expansion, damage to an internal structure, thermal runaway and the like of the power core are easily caused, so that the abnormal change of the internal pressure distribution of the battery pack is caused. Therefore, the pressure distribution of the battery pack is monitored and analyzed in real time, and the method has important significance for timely finding out abnormal states of the battery and guaranteeing safe operation of the system. The existing battery pack pressure distribution monitoring method mainly comprises the steps of collecting pressure data of a battery pack by arranging a pressure array, and judging whether the pressure distribution at the current moment is abnormal or not by setting a pressure threshold value or by counting a historical pressure distribution data range. However, under different operation conditions (such as different charge states, charge and discharge multiplying power, etc.), the pressure distribution rules of the battery pack have obvious differences, and the conventional monitoring method is difficult to reflect the normal pressure distribution characteristics under different operation conditions, so that the abnormal identification is often inaccurate easily. Therefore, how to improve the accuracy of monitoring the abnormal pressure distribution of the battery pack in real time is a problem to be solved. Disclosure of Invention In view of the above, the embodiment of the invention provides a method for dynamically monitoring the pressure distribution of a battery pack based on multi-sensor data fusion, so as to solve the problem of how to improve the accuracy of monitoring the abnormal pressure distribution of the battery pack in real time. The embodiment of the invention provides a battery pack pressure distribution dynamic monitoring method based on multi-sensor data fusion, which comprises the following steps: acquiring at least two working condition indexes for reflecting different operation working conditions of a target battery pack, and working condition data of each working condition index at each historical moment in a preset time period at and before the current moment; Aiming at any historical moment, according to the change trend stability of the working condition data of each working condition index in the local time range of any historical moment, acquiring an adaptive working condition window of any historical moment, wherein a target battery pack under each moment in the adaptive working condition window is judged to be in the same operating condition; Acquiring a self-adaptive working condition window of each historical moment and the current moment, and acquiring the operation working condition similarity of each historical moment and the current moment based on the data similarity of each working condition index in the self-adaptive working condition window of each historical moment and the current moment; according to the similarity of the operation working conditions of each historical moment and the current moment and the pressure of each monitoring point of the target battery pack at each historical moment, acquiring a pressure standard value of each monitoring point of the target battery pack, acquiring the pressure of each monitoring point of the target battery pack at the current moment, and according to the difference between the pressure of each monitoring point of the target battery pack at the current moment and the pressure standard value, monitoring the abnormal pressure distribution of the target battery pack at the current moment in real time. Preferably, the obtaining the adaptive working condition window at any historical moment according to the stability of the variation trend of the working condition data of each working condition index in the local time range at any historical moment includes: For any working condition index, forming an initial window of the any historical moment and two moments closest to the any historical moment, performing linear fitting on working condition data of the any working condition index at each moment in the initial window to obtain fitting data of the any working condition index at each moment in the initial window, and acquiring the fitting goodness of the any working condition index in the initial window according to the difference between