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

Abstract

This invention relates to the field of data processing technology, and in particular to a method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion. This method obtains an adaptive operating condition window for each moment based on the stability of the changing trends of various operating condition indicators of the target battery pack within a local time range. Based on the data similarity of various operating condition indicators within the adaptive operating condition window of each historical moment and the current moment, it obtains the operating condition similarity between each historical moment and the current moment. Then, combining the pressure at each monitoring point of the target battery pack at each historical moment, it obtains the pressure standard value at each monitoring point of the target battery pack. Based on the difference between the pressure at each monitoring point of the target battery pack at the current moment and the pressure standard value, it monitors for abnormal pressure distribution of the target battery pack at the current moment, thus improving the accuracy of real-time monitoring of abnormal pressure distribution in the battery pack.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion. Background Technology

[0002] With the widespread application of power batteries in new energy vehicles, energy storage systems, and other fields, the safety and stability of battery packs are receiving increasing attention in the monitoring of related system operations. When battery packs experience faults such as overcharging, overheating, or internal short circuits, they are prone to problems such as abnormal cell expansion, internal structural damage, and thermal runaway, leading to abnormal changes in the internal pressure distribution of the battery pack. Therefore, real-time monitoring and analysis of the pressure distribution of the battery pack is of great significance for timely detection of abnormal battery conditions and ensuring the safe operation of the system.

[0003] Existing battery pack pressure distribution monitoring methods primarily involve arranging pressure arrays to collect pressure data from the battery pack. They then determine whether there are anomalies in the current pressure distribution by setting pressure thresholds or by statistically analyzing historical pressure distribution data. However, the pressure distribution patterns of battery packs vary significantly under different operating conditions (such as different states of charge, charge / discharge rates, etc.). Traditional monitoring methods struggle to reflect the normal pressure distribution characteristics under different operating conditions, often leading to inaccurate anomaly identification.

[0004] Therefore, improving the accuracy of real-time monitoring of abnormal pressure distribution in battery packs has become an urgent problem to be solved. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion, in order to solve the problem of how to improve the accuracy of real-time monitoring of abnormal pressure distribution in battery packs.

[0006] This invention provides a method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion, the method comprising the following steps:

[0007] Acquire at least two operating condition indicators to reflect different operating conditions of the target battery pack, as well as operating condition data of each indicator at each historical moment in the current time and the preset time period before the current time.

[0008] For any historical moment, based on the stability of the changing trend of various operating condition data of the operating condition indicators within the local time range of the historical moment, an adaptive operating condition window for the historical moment is obtained. The target battery pack is determined to be the same operating condition at each moment within the adaptive operating condition window.

[0009] Obtain the adaptive operating condition window for each historical moment and the current moment. Based on the data similarity of various operating condition indicators within the adaptive operating condition window of each historical moment and the current moment, obtain the operating condition similarity between each historical moment and the current moment.

[0010] Based on the similarity of operating conditions between each historical moment and the current moment, and the pressure at each monitoring point of the target battery pack at each historical moment, the standard pressure value at each monitoring point of the target battery pack is obtained. The pressure at each monitoring point of the target battery pack at the current moment is obtained. Based on the difference between the pressure at each monitoring point of the target battery pack at the current moment and the standard pressure value, the abnormal pressure distribution of the target battery pack at the current moment is monitored in real time.

[0011] Preferably, obtaining the adaptive operating condition window for any historical time based on the stability of the changing trends of various operating condition indicators within a local time range of any historical time includes:

[0012] For any operating condition index, the historical time and the two times closest to the historical time are combined to form the initial window of the historical time. The operating condition data of the operating condition index at each time within the initial window are linearly fitted to obtain the fitted data of the operating condition index at each time within the initial window. Based on the difference between the operating condition data and the fitted data of the operating condition index at each time within the initial window, the goodness of fit of the operating condition index within the initial window is obtained and denoted as the initial goodness of fit.

[0013] The times adjacent to the start and end times of the initial window are obtained respectively and recorded as the adjacent times of the initial window. Any adjacent time is added to the initial window to obtain a new window for any historical time. The working condition data of any working condition index at each time in the new window are linearly fitted to obtain the goodness of fit of the working condition index in the new window, which is recorded as the new goodness of fit.

[0014] Subtracting the initial goodness of fit from the new goodness of fit yields the target difference. The ratio between the target difference and the initial goodness of fit is normalized to obtain the consistency of the data trend of any working condition index between any adjacent time and each time within the initial window.

[0015] Obtain a new window corresponding to another adjacent time point, and based on the goodness of fit of any working condition index within the new window corresponding to the other adjacent time point, obtain the degree of consistency of the data trend of any working condition index between the other adjacent time point and each time point within the initial window;

[0016] If the data trend consistency degree corresponding to any adjacent time point and the data trend consistency degree corresponding to another adjacent time point are both greater than the preset data trend consistency degree threshold, then the new window corresponding to the adjacent time point with the maximum data trend consistency degree is used as the initial window of any historical time point. If only one adjacent time point has a data trend consistency degree greater than the preset data trend consistency degree threshold, then the new window corresponding to the adjacent time point with a data trend consistency degree greater than the preset data trend consistency degree threshold is used as the initial window of any historical time point. The steps of obtaining the data trend consistency degree corresponding to the adjacent time points of the initial window are repeated until the data trend consistency degree corresponding to each adjacent time point of the initial window is less than or equal to the preset data trend consistency degree threshold. Then the initial window is used as the operating condition window of any operating condition indicator at any historical time point.

[0017] Obtain the operating condition windows for each operating condition indicator at any historical time. Record the window corresponding to the intersection of all operating condition windows as a common window. If the duration of the common window is greater than or equal to the preset minimum analysis duration, then the common window is used as the adaptive operating condition window for any historical time. If the duration of the common window is less than the preset minimum analysis duration, then the common window is evenly expanded to the preset minimum analysis duration in both left and right time directions to obtain the adaptive operating condition window for any historical time.

[0018] Preferably, obtaining the goodness of fit of any operating condition index within the initial window based on the difference between the operating condition data and the fitted data at each time point within the initial window includes:

[0019] Calculate the absolute value of the difference between the operating condition data and the fitted data at each time point within the initial window for any operating condition index, and use the negative of the average of all absolute differences as the independent variable of the natural exponential function to obtain the goodness of fit of any operating condition index within the initial window.

[0020] Preferably, the step of obtaining the operating condition similarity between each historical time and the current time based on the data similarity of various operating condition indicators within the adaptive operating condition window of each historical time and the current time includes:

[0021] Cluster the operating data of each operating condition indicator at the current time and at each historical time to obtain at least two clusters corresponding to each operating condition indicator. Based on the data stability within the clusters corresponding to each operating condition indicator and the difference between clusters, obtain the reflection weight of each operating condition indicator on different operating conditions of the target battery pack.

[0022] For any historical moment, the operating data of any operating condition index at each moment within the adaptive operating condition window of that historical moment are formed into a data sequence, which is denoted as the historical data sequence. The data sequence of the any operating condition index within the adaptive operating condition window at the current moment is obtained and denoted as the current data sequence. The DTW distance between the historical data sequence and the current data sequence is calculated. The reciprocal of the sum of the DTW distance and the preset constant is taken as the data similarity of the any operating condition index between the historical moment and the current moment.

[0023] The data similarity of each operating condition indicator between any historical time and the current time is obtained. Based on the reflection weight of each operating condition indicator, the data similarity of each operating condition indicator between any historical time and the current time is weighted and summed. The result of the weighted sum is normalized to obtain the operating condition similarity between any historical time and the current time.

[0024] Preferably, the step of obtaining the weights of each operating condition indicator in response to different operating conditions of the target battery pack based on the data stability within the cluster and the inter-cluster differences corresponding to each operating condition indicator includes:

[0025] For any operating condition index, obtain the cluster center of each cluster corresponding to the operating condition index, calculate the standard deviation of all cluster centers to obtain the inter-cluster standard deviation, and obtain the standard deviation of the data within each cluster corresponding to the operating condition index to obtain the intra-cluster standard deviation of each cluster corresponding to the operating condition index.

[0026] The maximum value among all intra-cluster standard deviations and inter-cluster standard deviations is denoted as the maximum standard deviation. The proportion of the inter-cluster standard deviation in the maximum standard deviation is normalized to obtain the first distinguishing factor.

[0027] Calculate the proportion of the intra-cluster standard deviation of each cluster corresponding to any of the operating conditions in the maximum standard deviation, and use the negative of the mean of the proportions corresponding to all clusters as the independent variable of the natural exponential function to obtain the second distinguishing factor.

[0028] Calculate the product between the first distinguishing factor and the second distinguishing factor to obtain the degree of distinguishing between different operating conditions of the target battery pack for any operating condition index;

[0029] The degree of differentiation of each operating condition indicator for different operating conditions of the target battery pack is obtained. The proportion of the degree of differentiation corresponding to any one operating condition indicator in the sum of all differentiation degrees is calculated to obtain the reflection weight of any one operating condition indicator for different operating conditions of the target battery pack.

[0030] Preferably, the step of obtaining the standard pressure value at each monitoring point of the target battery pack based on the similarity of operating conditions between each historical time and the current time, and the pressure at each monitoring point of the target battery pack at each historical time, includes:

[0031] The pressure at each monitoring point of the target battery pack at each historical time is obtained, and the standard deviation of the pressure at all monitoring points of the target battery pack at each historical time is calculated. The principle is to obtain the normal standard deviation range of all pressure standard deviations, and then remove historical moments where the pressure standard deviation exceeds the normal standard deviation range to obtain the target historical moment.

[0032] For any monitoring point of the target battery pack, the similarity of the operating conditions between each target historical time and the current time is used as a weight to calculate the weighted average of the pressure at any monitoring point under all target historical times, which is recorded as the standard pressure value at any monitoring point of the target battery pack.

[0033] Preferably, the step of real-time monitoring of pressure distribution anomalies in the target battery pack based on the difference between the pressure at each monitoring point of the target battery pack and the pressure standard value at the current moment includes:

[0034] For any monitoring point of the target battery pack, obtain the real-time pressure at the current time of the monitoring point, calculate the absolute value of the difference between the pressure standard value at the monitoring point and the real-time pressure, and take the proportion of the absolute value of the difference in the pressure standard value at the monitoring point as the pressure anomaly degree at the current time of the monitoring point.

[0035] The pressure anomaly level at any monitoring point at each historical time is obtained. Using a box plot, the upper limit of the anomaly level at any monitoring point at all historical times is obtained. If the pressure anomaly level at any monitoring point at the current time is greater than the upper limit of the anomaly level, a pressure anomaly alarm is triggered at any monitoring point of the target battery pack.

[0036] Preferably, the operating condition indicators include the current and state of charge of the target battery pack, as well as the temperature at each monitoring point of the target battery pack.

[0037] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0038] This invention analyzes the stability of data trends for various operating condition indicators within a local time range at each moment, obtaining adaptive operating condition windows for the current moment and each historical moment. This ensures consistent operating conditions for the target battery pack within each adaptive operating condition window while maximizing the acquisition of reference data for analysis. Furthermore, based on the similarity of operating condition data within each historical moment and the current moment's adaptive operating condition window, the similarity of the target battery pack's operating conditions between each historical moment and the current moment is analyzed. The results obtained from the adaptive operating condition window are more reliable than those from a fixed-size window. Using the similarity of the target battery pack's operating conditions between each historical moment and the current moment as weights, a normal pressure distribution model of the target battery pack under the current operating condition is constructed based on the pressure at each monitoring point of the target battery pack at each historical moment. This yields the standard pressure value at each monitoring point of the target battery pack, accurately reflecting the normal pressure distribution characteristics of the target battery pack under different operating conditions. Finally, based on the difference between the actual pressure at each monitoring point of the target battery pack at the current moment and the standard pressure value, real-time monitoring of abnormal pressure distribution of the target battery pack at the current moment is performed, improving the accuracy of real-time monitoring of abnormal pressure distribution of the battery pack. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a flowchart of a method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion, provided in Embodiment 1 of the present invention. Detailed Implementation

[0041] Embodiments of this disclosure are described in detail below, with examples of these embodiments illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this disclosure, and should not be construed as limiting it.

[0042] It should be noted that the terms "first," "second," etc., used in this disclosure and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure.

[0043] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0044] See Figure 1 This is a flowchart of a method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion, as provided in Embodiment 1 of the present invention. Figure 1 As shown, the method may include:

[0045] Step S101: Obtain at least two operating condition indicators to reflect different operating conditions of the target battery pack, as well as the operating condition data of each operating condition indicator at each historical moment in the current time and the preset time period before the current time.

[0046] Existing battery pack pressure distribution monitoring methods primarily involve arranging pressure arrays to collect pressure data from the battery pack. They then determine whether there are anomalies in the current pressure distribution by setting pressure thresholds or by statistically analyzing historical pressure distribution data. However, the pressure distribution patterns of battery packs vary significantly under different operating conditions (such as different states of charge, charge / discharge rates, etc.). Traditional monitoring methods struggle to reflect the normal pressure distribution characteristics under different operating conditions and cannot accurately determine whether pressure changes are caused by genuine anomalies or changes in operating conditions, thus affecting the accuracy of pressure distribution monitoring results.

[0047] This method analyzes the similarity between the battery pack's operating conditions at each historical moment and the current moment, and combines this with the pressure distribution at historical monitoring moments to construct a normal pressure distribution model of the battery pack under the current operating conditions. Based on the difference between the actual pressure data of the battery pack at the current moment and the normal pressure distribution model, abnormal pressure distribution of the battery pack can be identified.

[0048] Let any battery pack to be analyzed be designated as the target battery pack. First, obtain at least two operating condition indicators to reflect different operating conditions of the target battery pack. In this embodiment of the invention, the current, SOC (state of charge), and temperature at each monitoring point of the target battery pack are used as operating condition indicators (the temperature at each monitoring point is analyzed according to one operating condition indicator. Assuming there are 4 monitoring points, there are a total of 6 operating condition indicators). There is no limitation here, and the implementer can set it according to the specific scenario. The number of monitoring points of the target battery pack is determined according to the size of the target battery pack, such as arranging a 4×3 array. Then, the battery management system collects data on the target battery pack's current and SOC at each historical moment, both at the current moment and within the previous day (this is not limited; the implementer can set it according to the specific scenario, but it must ensure that the operating cycle of the target battery pack can be fully reflected within the preset time period). Simultaneously, temperature data is collected at each monitoring point of the target battery pack, both at the current moment and within the previous day. To facilitate comprehensive analysis of different operating conditions, a uniform collection frequency is set for all operating conditions, and the collected data for each operating condition are filtered, denoised, and normalized to obtain the operating condition data for each indicator at each moment. In this embodiment, the collection frequency is set to 10Hz, which is not limited; the implementer can set it according to the actual operating state of the target battery pack. Filtering, denoising, and normalization are existing technologies and will not be elaborated here. Since the monitoring involves the pressure distribution of the target battery pack, the pressure at each monitoring point of the target battery pack at each historical moment, both at the current moment and within the previous day, is also obtained through the pressure sensor array on top of the target battery pack (data after filtering, denoising, and normalization).

[0049] Step S102: For any historical moment, based on the stability of the changing trend of the operating condition data of various operating condition indicators within the local time range of the historical moment, obtain the adaptive operating condition window for the historical moment. The target battery pack is determined to be the same operating condition at each moment within the adaptive operating condition window.

[0050] Because the pressure distribution of the target battery pack varies under different operating conditions, in order to obtain the normal pressure distribution performance of the target battery pack at the current moment, it is necessary to increase the influence weight of historical pressure distribution data under similar operating conditions on the results. Therefore, after obtaining the operating condition data of each operating condition indicator reflecting different operating conditions of the target battery pack through step S101, the similarity between the operating conditions of the target battery pack at each historical moment and the current moment is first analyzed based on the operating condition data characteristics of each operating condition indicator at each historical moment within the day before the current moment.

[0051] Considering that operating data tends to be consistent under the same operating conditions, the similarity between historical and current operating data can be used to analyze the degree of similarity between the target battery pack's operating conditions at different times. However, considering that operating data at a single moment may be affected by instantaneous fluctuations, sensor noise, etc., similarity in operating data at a single moment does not necessarily indicate similarity between the corresponding historical moment and the current moment. Therefore, it is necessary to analyze the degree of similarity with the current operating conditions based on the data characteristics of each historical moment. However, if a fixed-size time window is set for each historical moment, the same time window may contain more than one operating condition, making it difficult to accurately determine the true operating condition at that historical moment. Considering that operating condition data usually changes stably according to a fixed trend under the same operating conditions, in this embodiment of the invention, the time range is gradually expanded based on the stability of the changing trend of various operating condition indicators between each time point and its adjacent time points to obtain an adaptive operating condition window for each historical time point and the current time point. The target battery pack is determined to be under the same operating condition at each time point within the adaptive operating condition window. Then, based on the similarity of the operating condition data of various operating condition indicators within the adaptive operating condition window of each historical time point and the current time point, the degree of similarity of the operating conditions of the target battery pack at each historical time point and the current time point is analyzed.

[0052] Taking the w-th historical moment as an example, the adaptive working condition window for the w-th historical moment is obtained as follows:

[0053] (1) Taking the a-th working condition index as an example, the w-th historical time and the two times closest to the w-th historical time are used to form the initial window of the w-th historical time. The least squares method is used to linearly fit the working condition data of the a-th working condition index at each time in the initial window to obtain the fitted data of the a-th working condition index at each time in the initial window. The least squares method is an existing technology and will not be described in detail here. The absolute value of the difference between the working condition data and the fitted data of the a-th working condition index at each time in the initial window is calculated. The negative of the average of all the absolute values ​​of the difference is used as the independent variable of the natural exponential function to obtain the goodness of fit of the a-th working condition index in the initial window, which is denoted as the initial goodness of fit.

[0054] In one embodiment, the formula for calculating the initial goodness of fit of the a-th operating condition index within the initial window at the w-th historical time is:

[0055]

[0056] in, This represents the initial goodness of fit of the a-th operating condition index within the initial window at the w-th historical time. This indicates the number of moments contained within the initial window. This represents the operating condition data of the 'a'-th operating condition indicator at the 'i'-th time within the initial window. This represents the fitted data of the working condition index a at the i-th time point within the initial window. Represents the absolute value symbol. This represents the natural exponential function.

[0057] It should be noted that, The smaller the value, the smaller the difference between the fitted data and the actual data within the initial window, the higher the goodness of fit of the initial window, and the more stable the trend of the working condition index a within the initial window. The larger the value, the greater the likelihood that the target battery pack is under the same operating condition within the initial window, which means the greater the likelihood that the operating condition of the target battery pack remains unchanged within the initial window.

[0058] (2) Obtain the times adjacent to the start and end times of the initial window respectively, and record them as the adjacent times of the initial window (there are at most two adjacent times and at least one). Add the j-th adjacent time to the initial window to obtain a new window for the w-th historical time. Perform linear fitting on the working condition data of the a-th working condition index at each time in the new window. According to the method of obtaining the initial goodness of fit in step (1), obtain the goodness of fit of the a-th working condition index in the new window, and record it as the new goodness of fit. express.

[0059] (3) Subtract the initial goodness of fit from the new goodness of fit to obtain the target difference. Normalize the ratio between the target difference and the initial goodness of fit to obtain the consistency of the data trend of the a-th working condition index between the j-th adjacent time and each time in the initial window.

[0060] In one embodiment, the formula for calculating the consistency of the data trend of the a-th operating condition index between the j-th adjacent time point and each time point within the initial window is as follows:

[0061]

[0062] in, This indicates the degree of consistency in the data trend of the a-th operating condition indicator between the j-th adjacent time point and each time point within the initial window. This represents the goodness of fit of the new window obtained after adding the a-th working condition index to the initial window at the j-th adjacent time, which is also the new goodness of fit. This represents the initial goodness of fit of the a-th operating condition index within the initial window at the w-th historical time. This represents the normalization function.

[0063] It should be noted that when hour, The larger the value, the greater the increase in the goodness of fit of the operating condition data of the a-th operating condition index after the j-th adjacent time point is added to the initial window. This indicates a more stable trend in the change of the a-th operating condition index data between the j-th adjacent time point and each time point within the initial window. The larger the value, the greater the likelihood that the target battery pack within the new window is under the same operating condition; when hour, The smaller, The larger the value, the greater the decrease in the goodness of fit of the operating condition index a after the j-th adjacent time point is added to the initial window. This indicates a change in the trend of the data change of the operating condition index a between the j-th adjacent time point and each time point within the initial window, and consequently... The smaller the value, the lower the probability that the target battery packs are in the same operating condition within the new window corresponding to the j-th adjacent time point; when When the time interval is reached, it indicates that the goodness of fit of the working condition data of the a-th working condition index did not change after the initial window was added at the j-th adjacent time.

[0064] (4) Similarly, obtain a new window corresponding to another adjacent time point, that is, add the other adjacent time point to the initial window to obtain a new window for the w-th historical time point, and obtain the data trend consistency degree of the a-th working condition index between the other adjacent time point and each time point in the initial window based on the goodness of fit of the a-th working condition index in the new window corresponding to the other adjacent time point; further, set a preset data trend consistency degree threshold, since The median value is taken as the threshold, that is, the preset data trend consistency threshold is set to 0.5. There is no restriction here. The implementer can set it according to the specific scenario. If the data trend consistency corresponding to the j-th adjacent time and the data trend consistency corresponding to the other adjacent time are both greater than 0.5, then the new window corresponding to the adjacent time corresponding to the maximum data trend consistency is taken as the initial window of the w-th historical time. If only one adjacent time corresponds to a data trend consistency greater than 0.5, then the new window corresponding to the adjacent time corresponding to the data trend consistency greater than 0.5 is taken as the initial window of the w-th historical time. Repeat the step (2) to obtain the data trend consistency of the adjacent time of the initial window until the data trend consistency of each adjacent time of the initial window is less than or equal to 0.5. Then take the initial window as the working condition window of the a-th working condition index at the w-th historical time.

[0065] (5) Similarly, the operating condition windows of each operating condition index at the w-th historical time are obtained. Since the data change amplitudes of different operating condition indices are different, the lengths of the operating condition windows of different operating condition indices at the w-th historical time are different. Therefore, in this embodiment of the invention, the window corresponding to the intersection of all operating condition windows is recorded as a common window. In order to ensure the statistical significance of subsequent analysis, if the duration corresponding to the common window is greater than or equal to the preset minimum analysis duration, the common window is used as the adaptive operating condition window at the w-th historical time. If the duration corresponding to the common window is less than the preset minimum analysis duration, the common window is evenly expanded to the left and right time directions (if it cannot be expanded in one direction, it is only expanded in the other direction) to the preset minimum analysis duration to obtain the adaptive operating condition window at the w-th historical time. Among them, for the setting of the preset minimum duration, since the data acquisition frequency in this embodiment of the invention is 10Hz, that is, 10 times of data are collected per second, in order to ensure the statistical significance of subsequent analysis, the number of data in the adaptive operating condition window is at least 10, so the preset minimum analysis duration is set to 1s. There is no restriction here, and the implementer can set it according to the data acquisition frequency.

[0066] Thus, the adaptive operating condition window for the w-th historical moment has been obtained.

[0067] Step S103: Obtain the adaptive operating condition window for each historical time and the current time. Based on the data similarity of various operating condition indicators within the adaptive operating condition window of each historical time and the current time, obtain the operating condition similarity between each historical time and the current time.

[0068] Following the method for obtaining the adaptive operating condition window of the w-th historical moment in step S102, the adaptive operating condition windows for the current moment and each historical moment within the day preceding the current moment are obtained. Further, based on the data similarity of various operating condition indicators within the adaptive operating condition windows of each historical moment and the current moment, the similarity of the target battery pack's operating conditions between each historical moment and the current moment is obtained, implemented as follows:

[0069] (1) Obtain the weights of each working condition index in response to different operating conditions of the target battery pack.

[0070] Because different operating condition indicators have different focuses in describing the operating status of the battery pack, and the data change characteristics of different operating condition indicators are different, the degree of differentiation of different operating condition indicators for different operating conditions of the target battery pack may vary. Therefore, when calculating the similarity between the operating conditions at historical time and the current time, it is first necessary to obtain the reflection weight of each operating condition indicator for different operating conditions of the target battery pack, so as to effectively reduce the impact of some normally fluctuating operating condition indicators on the operating condition similarity analysis results.

[0071] Taking the a-th operating condition indicator as an example, the specific method for obtaining the weight of the a-th operating condition indicator in relation to different operating conditions of the target battery pack is as follows:

[0072] Using the K-means clustering algorithm, the operating condition data of the a-th operating condition index at the current time and at each historical time are clustered to obtain at least two clusters corresponding to the a-th operating condition index. The K value in the K-means clustering algorithm can be determined by the elbow method. The K-means clustering algorithm and the elbow method are existing technologies and will not be described in detail here.

[0073] Since the higher the data stability within a cluster (i.e., the smaller the data difference within a cluster) and the more obvious the difference between clusters, it indicates that the working condition index a has a higher degree of differentiation for different working conditions. Therefore, we obtain the cluster center of each cluster corresponding to the working condition index a, calculate the standard deviation of all cluster centers to obtain the inter-cluster standard deviation, and obtain the standard deviation of the data within each cluster corresponding to the working condition index a to obtain the intra-cluster standard deviation of each cluster corresponding to the working condition index a.

[0074] The maximum value among all intra-cluster standard deviations and inter-cluster standard deviations is denoted as the maximum standard deviation. The proportion of the inter-cluster standard deviation in the maximum standard deviation is normalized to obtain the first distinguishing factor.

[0075] Calculate the proportion of the intra-cluster standard deviation of each cluster corresponding to the a-th working condition index in the maximum standard deviation, and use the negative of the mean of the proportions corresponding to all clusters as the independent variable of the natural exponential function to obtain the second distinguishing factor.

[0076] Calculate the product between the first distinguishing factor and the second distinguishing factor to obtain the degree of distinguishing between different operating conditions of the target battery pack by the a-th operating condition index;

[0077] Obtain the degree of differentiation of each operating condition indicator for different operating conditions of the target battery pack, calculate the proportion of the degree of differentiation corresponding to the a-th operating condition indicator in the sum of all differentiation degrees, and obtain the reflection weight of the a-th operating condition indicator for different operating conditions of the target battery pack.

[0078] In one embodiment, the formula for calculating the weighting of the operating condition index a in relation to different operating conditions of the target battery pack is as follows:

[0079]

[0080]

[0081] in, This indicates the weight of the operating condition indicator (item a) in relation to different operating conditions of the target battery pack. This indicates the degree to which the a-th operating condition indicator distinguishes different operating conditions of the target battery pack, and M represents the number of all operating condition indicators. Indicates the standard deviation between clusters. This represents the maximum standard deviation among all within-cluster standard deviations and between-cluster standard deviations, i.e., the maximum standard deviation. This represents the within-cluster standard deviation of the u-th cluster corresponding to the a-th operating condition indicator. This represents the number of clusters corresponding to the a-th operating condition indicator. Represents the normalization function. This represents the natural exponential function.

[0082] It should be noted that, The larger the value, the more significant the inter-cluster differences among the clusters corresponding to the a-th operating condition index, and the higher the degree of differentiation between different operating conditions indicated by the a-th operating condition index. The larger; The smaller the value, the higher the data stability within the cluster corresponding to the a-th operating condition indicator; that is, the smaller the data variability within the cluster, and the higher the degree to which the a-th operating condition indicator distinguishes between different operating conditions. The larger.

[0083] Similarly, obtain the weights of each operating condition indicator in response to different operating conditions of the target battery pack.

[0084] (2) Based on the data similarity of each operating condition index within the adaptive operating condition window between each historical time and the current time, and the weight of each operating condition index in response to different operating conditions of the target battery pack, the operating condition similarity between each historical time and the current time is obtained.

[0085] Specifically: Taking the w-th historical time and the a-th working condition indicator as an example, the working condition data of the a-th working condition indicator at each time point within the adaptive working condition window of the w-th historical time are composed into a data sequence, denoted as the historical data sequence. The data sequence of the a-th working condition indicator within the adaptive working condition window at the current time is obtained, denoted as the current data sequence. The DTW distance between the historical data sequence and the current data sequence is calculated. The reciprocal of the sum of the DTW distance and the preset constant is taken as the data similarity between the a-th working condition indicator at the w-th historical time and the current time. The calculation of the DTW distance is an existing technology and will not be elaborated here.

[0086] Obtain the data similarity between each operating condition indicator at the w-th historical time and the current time. Based on the reflection weight of each operating condition indicator, perform a weighted summation of the data similarity between each operating condition indicator at the w-th historical time and the current time. Normalize the result of the weighted summation to obtain the operating condition similarity between the w-th historical time and the current time.

[0087] In one implementation, the formula for calculating the similarity of the operating conditions between the w-th historical time and the current time is:

[0088]

[0089] in, This represents the similarity between the operating conditions at the w-th historical moment and the current moment. This indicates the weight of the operating condition indicator (item a) in relation to different operating conditions of the target battery pack. This represents the DTW distance between the historical data sequence of the 'a'-th operating condition index within the adaptive operating condition window at the 'w'-th historical time and the current data sequence within the current adaptive operating condition window. This represents a preset constant used to prevent the denominator from being zero. In this embodiment of the invention, it is set... There are no restrictions here; implementers can set them according to specific scenarios. M represents the number of all operating condition indicators. This represents the normalization function.

[0090] It should be noted that, The smaller the value, the more similar the data of the a-th operating condition indicator is to the adaptive operating condition window at the w-th historical time, indicating that the operating condition of the target battery pack at the w-th historical time is closer to that at the current time, and thus... The larger; The larger the value, the more significant the effect of differentiating the target battery pack under different operating conditions based on the operating condition index of item a. The greater the credibility, the more... The larger.

[0091] Similarly, obtain the similarity of the operating conditions between each historical moment and the current moment.

[0092] Step S104: Based on the similarity of the operating conditions between each historical time and the current time, and the pressure at each monitoring point of the target battery pack at each historical time, obtain the pressure standard value at each monitoring point of the target battery pack, obtain the pressure at each monitoring point of the target battery pack at the current time, and monitor the pressure distribution anomaly of the target battery pack at the current time in real time based on the difference between the pressure at each monitoring point of the target battery pack at the current time and the pressure standard value.

[0093] After obtaining the similarity of operating conditions between each historical moment and the current moment through step S103, the standard pressure value at each monitoring point of the target battery pack is obtained by combining the pressure at each monitoring point of the target battery pack at each historical moment. Specifically:

[0094] The pressure at each monitoring point of the target battery pack at each historical time is obtained. Considering that battery pack pressure anomalies mostly manifest as localized pressure anomalies, in order to minimize the impact of historical anomaly data, this embodiment of the invention calculates the standard deviation of pressure at all monitoring points of the target battery pack at each historical time, and utilizes... The principle is to obtain the normal standard deviation range for all pressure standard deviations. The principle is based on existing technology and will not be elaborated here. Historical moments where the pressure standard deviation exceeds the normal standard deviation range are removed to obtain the target historical moment.

[0095] Taking the qth monitoring point of the target battery pack as an example, the similarity of the operating conditions between each target historical time and the current time is used as a weight to calculate the weighted average of the pressure at the qth monitoring point under all target historical times, which is recorded as the standard pressure value at the qth monitoring point of the target battery pack.

[0096] In one embodiment, the formula for calculating the standard pressure value at the q-th monitoring point is:

[0097]

[0098] in, This represents the standard pressure value at the q-th monitoring point. This represents the similarity between the historical operating conditions of the t-th target and its current operating condition. Let T represent the pressure at the q-th monitoring point of the target battery pack at the t-th target historical moment, and let T represent the number of all target historical moments.

[0099] Similarly, the standard pressure value at each monitoring point of the target battery pack is obtained. Furthermore, based on the difference between the real-time pressure at each monitoring point of the target battery pack and the standard pressure value at the current moment, the degree of pressure anomaly at each monitoring point of the target battery pack at the current moment is obtained, specifically:

[0100] Taking the qth monitoring point of the target battery pack as an example, the real-time pressure at the qth monitoring point at the current moment is obtained, the absolute value of the difference between the pressure standard value at the qth monitoring point and the real-time pressure is calculated, and the proportion of the absolute value of the difference in the pressure standard value at the qth monitoring point is taken as the pressure anomaly degree at the qth monitoring point at the current moment.

[0101] In one embodiment, the formula for calculating the degree of pressure anomaly at the q-th monitoring point is:

[0102]

[0103] in, This indicates the degree of pressure anomaly at the q-th monitoring point. This represents the real-time pressure at the q-th monitoring point at the current moment. This represents the standard pressure value at the q-th monitoring point. Represents the absolute value symbol.

[0104] It should be noted that, The smaller the value, the closer the real-time pressure at the q-th monitoring point of the target battery pack is to the pressure standard value, the more normal the pressure at the q-th monitoring point of the target battery pack at the current moment, and the smaller the degree of pressure abnormality at the q-th monitoring point of the target battery pack at the current moment. The smaller.

[0105] Similarly, the degree of pressure anomaly at each monitoring point of the target battery pack at the current moment is obtained. Furthermore, based on the degree of pressure anomaly at each monitoring point of the target battery pack at the current moment, the pressure distribution anomaly of the target battery pack at the current moment is monitored in real time. Specifically:

[0106] Since the proportion of abnormal pressure data is relatively small compared to normal data, and the difference in the degree of abnormal pressure is large, the pressure abnormality at each monitoring point of the target battery pack at each historical moment is obtained in the above manner. The upper limit of the pressure abnormality at each monitoring point of the target battery pack is obtained using the box plot method, which is denoted as the abnormality upper limit. The box plot method is existing technology and will not be elaborated here. If the pressure abnormality at the q-th monitoring point of the target battery pack at the current moment is greater than the abnormality upper limit at the q-th monitoring point, it is considered that the q-th monitoring point of the target battery pack has an abnormality at the current moment. At this time, it is necessary to issue a pressure abnormality alarm for the target battery pack and report the location of the monitoring point where the pressure abnormality occurred.

[0107] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion, characterized in that, The method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion includes: Acquire at least two operating condition indicators to reflect different operating conditions of the target battery pack, as well as operating condition data of each indicator at each historical moment in the current time and the preset time period before the current time. For any historical moment, based on the stability of the changing trend of various operating condition data of the operating condition indicators within the local time range of the historical moment, an adaptive operating condition window for the historical moment is obtained. The target battery pack is determined to be the same operating condition at each moment within the adaptive operating condition window. Obtain the adaptive operating condition window for each historical moment and the current moment. Based on the data similarity of various operating condition indicators within the adaptive operating condition window of each historical moment and the current moment, obtain the operating condition similarity between each historical moment and the current moment. Based on the similarity of the operating conditions between each historical moment and the current moment, and the pressure at each monitoring point of the target battery pack at each historical moment, the standard pressure value at each monitoring point of the target battery pack is obtained, the pressure at each monitoring point of the target battery pack at the current moment is obtained, and the pressure distribution anomaly of the target battery pack at the current moment is monitored in real time based on the difference between the pressure at each monitoring point of the target battery pack at the current moment and the standard pressure value. The process of obtaining the operational condition similarity between each historical time and the current time based on the data similarity of various operational indicators within the adaptive operational condition window includes: Cluster the operating data of each operating condition indicator at the current time and at each historical time to obtain at least two clusters corresponding to each operating condition indicator. Based on the data stability within the clusters corresponding to each operating condition indicator and the difference between clusters, obtain the reflection weight of each operating condition indicator on different operating conditions of the target battery pack. For any historical moment, the operating data of any operating condition index at each moment within the adaptive operating condition window of that historical moment are formed into a data sequence, which is denoted as the historical data sequence. The data sequence of the any operating condition index within the adaptive operating condition window at the current moment is obtained and denoted as the current data sequence. The DTW distance between the historical data sequence and the current data sequence is calculated. The reciprocal of the sum of the DTW distance and the preset constant is taken as the data similarity of the any operating condition index between the historical moment and the current moment. The data similarity of each operating condition indicator between any historical time and the current time is obtained. Based on the reflection weight of each operating condition indicator, the data similarity of each operating condition indicator between any historical time and the current time is weighted and summed. The result of the weighted sum is normalized to obtain the operating condition similarity between any historical time and the current time.

2. The method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion according to claim 1, characterized in that, The step of obtaining the adaptive operating condition window for any historical moment based on the stability of the changing trends of various operating condition indicators within a local time range at any historical moment includes: For any operating condition index, the historical time and the two times closest to the historical time are combined to form the initial window of the historical time. The operating condition data of the operating condition index at each time within the initial window are linearly fitted to obtain the fitted data of the operating condition index at each time within the initial window. Based on the difference between the operating condition data and the fitted data of the operating condition index at each time within the initial window, the goodness of fit of the operating condition index within the initial window is obtained and denoted as the initial goodness of fit. The times adjacent to the start and end times of the initial window are obtained respectively and recorded as the adjacent times of the initial window. Any adjacent time is added to the initial window to obtain a new window for any historical time. The working condition data of any working condition index at each time in the new window are linearly fitted to obtain the goodness of fit of the working condition index in the new window, which is recorded as the new goodness of fit. Subtracting the initial goodness of fit from the new goodness of fit yields the target difference. The ratio between the target difference and the initial goodness of fit is normalized to obtain the consistency of the data trend of any working condition index between any adjacent time and each time within the initial window. Obtain a new window corresponding to another adjacent time point, and based on the goodness of fit of any working condition index in the new window corresponding to the other adjacent time point, obtain the degree of consistency of the data trend of any working condition index between the other adjacent time point and each time point in the initial window; If the data trend consistency degree corresponding to any adjacent time point and the data trend consistency degree corresponding to another adjacent time point are both greater than the preset data trend consistency degree threshold, then the new window corresponding to the adjacent time point with the maximum data trend consistency degree is used as the initial window of any historical time point. If only one adjacent time point has a data trend consistency degree greater than the preset data trend consistency degree threshold, then the new window corresponding to the adjacent time point with a data trend consistency degree greater than the preset data trend consistency degree threshold is used as the initial window of any historical time point. The steps of obtaining the data trend consistency degree corresponding to the adjacent time points of the initial window are repeated until the data trend consistency degree corresponding to each adjacent time point of the initial window is less than or equal to the preset data trend consistency degree threshold. Then the initial window is used as the operating condition window of any operating condition indicator at any historical time point. Obtain the operating condition windows for each operating condition indicator at any historical time. Record the window corresponding to the intersection of all operating condition windows as a common window. If the duration of the common window is greater than or equal to the preset minimum analysis duration, then the common window is used as the adaptive operating condition window for any historical time. If the duration of the common window is less than the preset minimum analysis duration, then the common window is evenly expanded to the preset minimum analysis duration in both left and right time directions to obtain the adaptive operating condition window for any historical time.

3. The method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion according to claim 2, characterized in that, The step of obtaining the goodness of fit of any operating condition index within the initial window based on the difference between the operating condition data and the fitted data at each time point within the initial window includes: Calculate the absolute value of the difference between the operating condition data and the fitted data at each time point within the initial window for any operating condition index, and use the negative of the average of all absolute differences as the independent variable of the natural exponential function to obtain the goodness of fit of any operating condition index within the initial window.

4. The method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion according to claim 1, characterized in that, The step of obtaining the weights of each operating condition indicator in response to different operating conditions of the target battery pack based on the data stability within the cluster and the inter-cluster differences corresponding to each operating condition indicator includes: For any operating condition index, obtain the cluster center of each cluster corresponding to the operating condition index, calculate the standard deviation of all cluster centers to obtain the inter-cluster standard deviation, and obtain the standard deviation of the data within each cluster corresponding to the operating condition index to obtain the intra-cluster standard deviation of each cluster corresponding to the operating condition index. The maximum value among all intra-cluster standard deviations and inter-cluster standard deviations is denoted as the maximum standard deviation. The proportion of the inter-cluster standard deviation in the maximum standard deviation is normalized to obtain the first distinguishing factor. Calculate the proportion of the intra-cluster standard deviation of each cluster corresponding to any of the operating conditions in the maximum standard deviation, and use the negative of the mean of the proportions corresponding to all clusters as the independent variable of the natural exponential function to obtain the second distinguishing factor. Calculate the product between the first distinguishing factor and the second distinguishing factor to obtain the degree of distinguishing between different operating conditions of the target battery pack for any operating condition index; The degree of differentiation of each operating condition indicator for different operating conditions of the target battery pack is obtained. The proportion of the degree of differentiation corresponding to any one operating condition indicator in the sum of all differentiation degrees is calculated to obtain the reflection weight of any one operating condition indicator for different operating conditions of the target battery pack.

5. The method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion according to claim 1, characterized in that, The step of obtaining the standard pressure value at each monitoring point of the target battery pack based on the similarity of operating conditions between each historical moment and the current moment, and the pressure at each monitoring point of the target battery pack at each historical moment, includes: The pressure at each monitoring point of the target battery pack at each historical time is obtained, and the standard deviation of the pressure at all monitoring points of the target battery pack at each historical time is calculated. The principle is to obtain the normal standard deviation range of all pressure standard deviations, and then remove historical moments where the pressure standard deviation exceeds the normal standard deviation range to obtain the target historical moment. For any monitoring point of the target battery pack, the similarity of the operating conditions between each target historical time and the current time is used as a weight to calculate the weighted average of the pressure at any monitoring point under all target historical times, which is recorded as the standard pressure value at any monitoring point of the target battery pack.

6. The method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion according to claim 1, characterized in that, The method of real-time monitoring of pressure distribution anomalies in the target battery pack based on the difference between the pressure at each monitoring point of the target battery pack and the pressure standard value at the current moment includes: For any monitoring point of the target battery pack, obtain the real-time pressure at the current time of the monitoring point, calculate the absolute value of the difference between the pressure standard value at the monitoring point and the real-time pressure, and take the proportion of the absolute value of the difference in the pressure standard value at the monitoring point as the pressure anomaly degree at the current time of the monitoring point. The pressure anomaly level at any monitoring point at each historical time is obtained. Using a box plot, the upper limit of the anomaly level at any monitoring point at all historical times is obtained. If the pressure anomaly level at any monitoring point at the current time is greater than the upper limit of the anomaly level, a pressure anomaly alarm is triggered at any monitoring point of the target battery pack.

7. The method for dynamic monitoring of battery pack pressure distribution based on multi-sensor data fusion according to claim 1, characterized in that, The operating parameters include the current and state of charge of the target battery pack, as well as the temperature at each monitoring point of the target battery pack.

Images