A method and system for monitoring the operation of a lithium battery power module

Abstract

The application relates to the technical field of data processing, in particular to a lithium battery power module operation monitoring method and system. The method comprises the following steps: constructing a fault propagation chain of each fault type; acquiring an abnormal value sequence corresponding to each parameter real-time data, and monitoring an abnormal rising trend of each parameter in real time to construct a real-time propagation chain; comparing the real-time propagation chain and the fault propagation chain of each fault type to obtain the probability of each fault type at any moment. Through the technical scheme, the change of each parameter before the fault occurs can be considered, and the operation of the lithium battery power module can be accurately acquired.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a lithium battery power module operation monitoring method and system. BACKGROUND

[0002] Lithium batteries are a kind of rechargeable batteries, which are favored by users because of their relatively light weight, high energy density, fast charging speed and no memory effect, and are used as power modules in new energy vehicles and electric bicycles. However, due to the characteristics of high active chemical composition of lithium batteries, triggered by misuse of electricity, heat, and machinery, etc., it is easy to induce battery thermal runaway and fire. Once the early warning is not timely, it will pose a serious threat to the safety of personnel and surrounding vehicles, and it is necessary to monitor the operation process of the lithium battery power module.

[0003] At present, the patent application file with the application publication number CN115792629A discloses a lithium battery energy storage alarm monitoring system and method, wherein the method comprises: collecting the operation data of the lithium battery energy storage process in real time, uploading the leakage detection data, temperature detection data, environmental humidity detection data and charge-discharge detection data of the lithium battery energy storage process to the operation control platform; the operation control platform compares the leakage detection data, temperature detection data, environmental humidity detection data and charge-discharge detection data with the corresponding pre-set threshold parameters respectively, and when any one of the leakage detection data, temperature detection data, environmental humidity detection data and charge-discharge detection data exceeds the corresponding pre-set threshold parameter, evaluates the operation of the lithium battery by using a comprehensive evaluation model to obtain an evaluation score; compare the evaluation score with the preset evaluation score threshold, and when the evaluation score is lower than the preset evaluation score threshold, determine that the current lithium battery operation is not normal; remotely control the lithium battery operation and alarm according to whether the current lithium battery operation is normal.

[0004] The above method compares the detection results of each parameter in the lithium battery energy storage process with the threshold parameters to comprehensively evaluate the operation process of the lithium battery, and then judges whether the current lithium battery operation is normal. However, when the lithium battery fails, not all parameter detection results will exceed the threshold parameter, often a parameter will first appear abnormal, and then a series of parameter changes will occur. The above method ignores the changes of each parameter before the failure, simply compares the detection results of each parameter with the threshold parameters, and cannot accurately obtain the operation of the lithium battery power module. SUMMARY

[0005] In order to solve the technical problem that the operation of the lithium battery power module cannot be accurately obtained, the present application provides a lithium battery power module operation monitoring method and system, which can consider the changes of each parameter before the failure, and accurately obtain the operation of the lithium battery power module.

[0006] In a first aspect, the application provides a lithium battery power module operation monitoring method, the monitoring method comprising: constructing a fault propagation chain of each fault type, comprising: performing anomaly detection on the time series of each parameter in the historical data of any fault type to obtain an abnormal value sequence of each parameter; calculating the abnormal rising trend of each time point in any abnormal value sequence; connecting each parameter in the order of the abnormal time point of each parameter from small to large, the abnormal time point being the time point at which the abnormal rising trend first exceeds a threshold; taking the abnormal interval and the trend increment as the edge weight of each parameter except the first parameter to obtain the fault propagation chain of the fault type; the abnormal interval and the trend increment are the interval between the abnormal time point of any parameter and the previous adjacent parameter, and the difference between the abnormal rising trends of the abnormal time points; obtaining the abnormal value sequence corresponding to the real-time data of each parameter, and real-time monitoring the abnormal rising trend of each parameter to construct a real-time propagation chain; comparing the real-time propagation chain with the fault propagation chain of each fault type to obtain the probability of each fault type at any time.

[0007] The fault propagation chain of each fault type is constructed through historical data, which can accurately capture the evolution process of the abnormal trend of each parameter before the fault occurs; during the operation of the lithium battery power module, the abnormal rising trend of each parameter is monitored in real time to construct a real-time propagation chain, which can reflect the time sequence and intensity change of the abnormal parameters during the operation; and the real-time propagation chain is matched with the fault propagation chain of each fault type to accurately determine the probability of each fault type at any time.

[0008] Preferably, the parameters include voltage, current, internal resistance, battery temperature and state of charge.

[0009] The core performance parameters of lithium batteries such as voltage, current, internal resistance, temperature and SOC are covered to ensure that the propagation chain can comprehensively reflect the electrical and thermal runaway fault characteristics.

[0010] Preferably, the anomaly detection comprises: dividing the time series of any parameter into a plurality of sub-sequences according to a preset window, and obtaining the abnormal values of each sub-sequence using the LOF algorithm, the abnormal values of each sub-sequence constituting the abnormal value sequence of the parameter.

[0011] Preferably, the time point The abnormal rising trend is:

[0012] ; The neighborhood interval of the time point ; And is a set of numerical pairs, which are the neighborhood interval of the time point , the time point and the time point Parameter value of Indicates the acquisition time The neighborhood interval of All value pairs of , is the mean value function.

[0013] In the neighborhood Calculate the average slope of all data pairs, suppress instantaneous noise interference, and accurately calculate the time abnormal upward trend.

[0014] Preferably, the construction of the fault propagation chain for each fault type also includes: counting the number of occurrences of each parameter sequence in the fault propagation chains corresponding to multiple historical data, determining the parameter sequence with the largest number of occurrences, and then calculating the average value of the edge weights of each parameter in all fault propagation chains corresponding to the parameter sequence to obtain the final fault propagation chain for the fault type.

[0015] The high-frequency parameter sequences of multiple historical data are counted to ensure the accuracy of fault propagation; edge weights are averaged to reduce the deviation of single samples and improve the generalization ability of the propagation chain.

[0016] Preferably, the real-time monitoring of the abnormal rising trend of each parameter to construct a real-time propagation chain includes: in response to the abnormal moment of any parameter at the current moment, constructing a real-time propagation chain with the parameter as the starting end at the current moment; after the current moment, if any other parameter other than the parameter has an abnormal moment without exceeding the maximum waiting time, connecting the other parameters to the terminal end of the real-time propagation chain, and calculating the edge weights of the other parameters, until all parameters are traversed to complete the construction of the real-time propagation chain.

[0017] The construction is started only when an anomaly is detected, saving computing power; the propagation chain is dynamically expanded according to the actual time sequence of the anomaly occurrence, realizing the efficient construction of the fault propagation chain.

[0018] Preferably, the time Fault type Probability The acquisition steps include: responding to the moment There is no established real-time propagation chain, the fault type Probability is 0, otherwise, calculate the time Each real-time propagation chain and fault type The similarity between fault propagation chains, taking the maximum similarity as the moment Fault type Probability .

[0019] Get the real-time transmission chain built at any time, and only participate in the probability calculation when the real-time transmission chain is complete, to ensure that the time Fault type Probability accuracy to avoid false positives.

[0020] Preferably, real-time transmission chain and fault type Fault propagation chain The similarity between for: ; Real-time communication chain and fault propagation chains length; and Fault propagation chain and real-time communication chain Middle parameters, if , ,like , 0; and Fault propagation chain and real-time communication chain Middle The edge weight after parameter standardization, for The L1 norm of .

[0021] Preferably, before obtaining the abnormal value sequence of each parameter, the monitoring method further comprises: performing normalization processing on the time series of each parameter.

[0022] In the second aspect of the present application, a lithium battery power module operation monitoring system is also provided, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a lithium battery power module operation monitoring method according to the first aspect of the present application is implemented.

[0023] The technical solution of this application has the following beneficial technical effects:

[0024] The fault propagation chain of each fault type is constructed through historical data, which reflects the time sequence and intensity change of abnormal parameters when the fault occurs, and can accurately capture the evolution process of abnormal trend of each parameter before the fault occurs; during the working process of the lithium battery power module, the abnormal rising trend of each parameter is monitored in real time to construct a real-time propagation chain, which can reflect the time sequence and intensity change of abnormal parameters during the working process; and the real-time propagation chain is matched with the fault propagation chain of each fault type, the evolution process of the abnormal trend of each parameter before the fault occurs is comprehensively considered, the probability of each fault type at any time is accurately judged, and then the running condition of the lithium battery power module is accurately obtained, and false detection is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 is a flowchart of a lithium battery power module running monitoring method according to an embodiment of the application.

[0026] Figure 2 is a schematic diagram of a fault propagation chain of a fault type corresponding to Table 1 according to an embodiment of the application.

[0027] Figure 3 is a structural block diagram of a lithium battery power module running monitoring system according to an embodiment of the application. DETAILED DESCRIPTION

[0028] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are some of the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the application.

[0029] According to a first aspect of the application, the application provides a lithium battery power module running monitoring method. Figure 1 is a flowchart of a lithium battery power module running monitoring method according to an embodiment of the application. As shown in Figure 1 , the lithium battery power module running monitoring method comprises steps S101 to S103, which will be described in detail below.

[0030] S101, constructing a fault propagation chain of each fault type.

[0031] In one embodiment, the lithium battery power module performs multiple charging and discharging processes during the working process, and various fault types such as short circuit, overcharge or loose connector may occur during the charging and discharging process. In any charging and discharging process, if the lithium battery power module fails, the time sequence of each parameter of the charging and discharging process is collected as historical data of the corresponding fault type.

[0032] Exemplarily, in a scenario of starting a new energy vehicle by using a lithium battery power module, if a connector loosening fault occurs in the lithium battery power module in a starting process, the time sequence of each parameter in this starting process is taken as historical data of the fault type "connector loosening".

[0033] The parameters include voltage, current, internal resistance, battery temperature and state of charge (SOC).

[0034] In one embodiment, after obtaining the historical data of each fault type, a fault propagation chain of each fault type can be constructed according to the historical data of each fault type. Specifically, constructing the fault propagation chain of each fault type includes: performing anomaly detection on the time sequence of each parameter in the historical data of any fault type to obtain an abnormal value sequence of each parameter; calculating an abnormal rising trend at each time in any abnormal value sequence; connecting each parameter in the order of abnormal time of each parameter from small to large, the abnormal time being the time when the abnormal rising trend is first greater than a threshold, and taking the abnormal interval and the trend increment as the edge weight of each parameter except the first parameter to obtain the fault propagation chain of the fault type, the abnormal interval being the interval of abnormal time between any parameter and the adjacent parameter, and the trend increment being the difference of abnormal rising trend between any parameter and the adjacent parameter.

[0035] The anomaly detection includes: dividing the time sequence of any parameter into a plurality of subsequences according to a preset window, and obtaining abnormal values of each subsequence by using the LOF algorithm, the abnormal values of each subsequence constituting the abnormal value sequence of the parameter.

[0036] Exemplarily, the size of the preset window is 5x1, that is, one subsequence contains 5 parameter values, and the preset window slides on the time sequence with a step of 1 until all values of the time sequence are traversed, and then the abnormal values of each subsequence are arranged in the order of the subsequences to obtain the abnormal value sequence. If the length of the time sequence is 50, when the size of the preset window is 5x1, the abnormal value sequence includes 46 abnormal values.

[0037] The time The abnormal rising trend is:

[0038] ; The neighborhood interval of the time ; and are a group of numerical pairs, which are the neighborhood interval of the time and the time in the neighborhood interval of the time Parameter value of Indicates the acquisition time The neighborhood interval of All value pairs of , is the mean value function. The value of is 5.

[0039] It should be noted that in order to avoid the influence of noise on the outlier sequence and make the outlier sequence truly reflect the operation status of the lithium battery power module, before obtaining the outlier sequence, the time series of each parameter in the historical data is preprocessed, and the preprocessing includes normalization and denoising; the normalization is used to eliminate the influence of the dimension on the abnormal rising trend of each parameter, and the denoising is used to eliminate the influence of noise on the outlier sequence.

[0040] It is understandable that when the lithium battery power module operates normally, the abnormal values ​​at each moment should be consistent in the region, that is, at each moment Abnormal upward trend The value of is close to 0; when When it is greater than 0, it means that the abnormal value is at time There is an upward trend in the neighborhood of The larger it is, the more obvious the upward trend is, which means the more obvious the abnormality of the parameter is.

[0041] In one embodiment, the abnormal upward trend of a parameter at any moment is calculated. When the abnormal upward trend of a parameter exceeds a threshold, the parameter is considered abnormal at that moment. The moment when the parameter first becomes abnormal is used as the abnormal moment. The parameter at the first abnormal moment is used as the starting parameter, that is, the initial parameter is considered to be the trigger of the fault type. The threshold value is 0.5.

[0042] After connecting each parameter in ascending order of their anomaly moments, the anomaly interval and trend increment are calculated for each parameter. These are used as edge weights to construct the fault propagation chain for the corresponding fault type. The anomaly interval of a parameter characterizes the lag in the occurrence of an anomaly in that parameter after the occurrence of an anomaly in the previous adjacent parameter. For example, an anomaly interval of 3 indicates that the anomaly occurred three moments after the occurrence of an anomaly in the previous adjacent parameter. The trend increment is the difference in the anomaly upward trend between any parameter and the previous adjacent parameter at the time of the anomaly, reflecting the relative relationship between the anomaly upward trend of any parameter and the previous adjacent parameter in the fault propagation chain.

[0043] It can be understood that the edge weight of an edge includes two values; if there are two other parameters with the same lag time, the other parameter with a larger trend increment can be placed in front.

[0044] In one example, the number of parameters is 5, the abnormal time of each parameter, and the abnormal rising trend of the abnormal time are shown in Table 1; and a fault propagation chain corresponding to the fault type can be constructed according to Table 1, please refer to Figure 2 FIG. 1 is a schematic diagram of the fault propagation chain corresponding to the fault type of Table 1 in the embodiment of the present application.

[0045] Table 1 shows the lag time and relative trend of other parameters.

[0046]

[0047] In this way, the construction of the fault propagation chain of each fault type is completed, and the evolution process of all parameter abnormal values of each fault type can be intuitively understood according to the fault propagation chain of each fault type, thereby providing a data basis for subsequent implementation of lithium battery power module operation monitoring.

[0048] It can be understood that one piece of historical data of one fault type can construct one fault propagation chain; and multiple pieces of historical data can construct multiple fault propagation chains. Considering that there are noises and the like in historical data, random selection of one fault propagation chain will inevitably lead to errors of the fault propagation chain. Therefore, the construction of the fault propagation chain of each fault type further includes: in the fault propagation chains corresponding to multiple pieces of historical data, the occurrence frequency of each parameter sequence is counted, the parameter sequence with the highest occurrence frequency is determined, the average value of the edge weight of each parameter in all fault propagation chains corresponding to the parameter sequence is calculated, and finally the fault propagation chain of the fault type is obtained. The accuracy of the fault propagation chain of each fault type is ensured.

[0049] In S102, the abnormal value sequence corresponding to the real-time data of each parameter is obtained, and the abnormal rising trend of each parameter is monitored in real time to construct a real-time propagation chain.

[0050] In one embodiment, during the operation of the lithium battery power module, the real-time data of each parameter is collected, the real-time data is a time sequence; the abnormal value of each time in the real-time data is obtained to obtain the abnormal value sequence of the real-time data of each parameter; and the abnormal rising trend of the real-time data at each time is calculated.

[0051] During the real-time operation of the lithium battery power module, the real-time monitoring of the abnormal rising trend of each parameter to construct a real-time propagation chain includes: in response to the abnormal time of an arbitrary parameter at the current time, a real-time propagation chain with the parameter as the starting end is constructed at the current time; after the current time, if an arbitrary other parameter appears an abnormal time without exceeding the maximum waiting time, the other parameter is connected to the termination end of the real-time propagation chain, and the edge weight of the other parameter is calculated; until all parameters are traversed, the construction of the real-time propagation chain is completed.

[0052] It is understandable that during the real-time operation of the lithium battery power module, if all parameters do not detect an abnormal moment, the construction of the real-time propagation chain will not be executed; whenever an abnormal moment is detected, the construction of the real-time propagation chain will be executed starting from the abnormal moment, with the parameters corresponding to the abnormal moment as the starting end; the maximum waiting time is a preset multiple of the maximum interval of abnormal moments in all fault propagation chains, and the preset multiple is greater than 1. In the embodiment of the present application, the value of the preset multiple is 1.5. For example, if the moment If the monitoring parameters 1 and 3 are abnormal, then at time Parameter 1 and parameter 3 are used as the starting points to build a real-time propagation chain, that is, at time There are 2 real-time transmission chains being built; if the moment If the parameter 2 is detected as abnormal at time The real-time transmission chain is constructed again with parameter 2 as the starting point, that is, at time There is 1 real-time transmission chain being built.

[0053] Among them, in a real-time propagation chain, other parameters except the starting segment correspond to a set of edge weights, which also include abnormal intervals and trend increments; the abnormal intervals of other parameters in the real-time propagation chain are the intervals between the abnormal moments of the other parameters and the previous adjacent parameters in the real-time propagation chain; the trend increments of other parameters in the real-time propagation chain are the differences in the abnormal rising trends between the other parameters and the previous adjacent parameters in the real-time propagation chain at the abnormal moments.

[0054] In this way, multiple real-time propagation chains will be constructed during the real-time operation of the lithium battery power module. It should be noted that there are incomplete real-time propagation chains. If a real-time propagation chain is incomplete, that is, the real-time propagation chain does not contain all parameters, the real-time propagation chain will be discarded and only the complete real-time propagation chain will be retained; each retained real-time propagation chain contains all parameters. At this time, the length of each real-time propagation chain is the same, and the length of the real-time propagation chain is equal to the fault propagation chain of each fault type.

[0055] S103 , comparing the real-time propagation chain with the fault propagation chain of each fault type to obtain the probability of each fault type at any time.

[0056] In one embodiment, during the real-time operation of the lithium battery power module, multiple real-time propagation chains may be constructed at a moment, or there may be no real-time propagation chain; when there is no real-time propagation chain at a moment, it means that the probability of each fault type at that moment is 0, and the lithium battery power module is in a normal operating state.

[0057] Specific, moment Fault type Probability The acquisition steps include: responding to the moment There is no completed real-time propagation chain, the probability of fault type is 0, otherwise, the time is calculated The similarity between each real-time propagation chain and fault type The maximum similarity between the fault propagation chain is taken as the time Fault type The probability of .

[0058] Wherein, the similarity between the real-time propagation chain and the fault type The similarity between the fault propagation chain is: ; is the length of the real-time propagation chain and the fault propagation chain ; and are the first parameters in the fault propagation chain and the real-time propagation chain , if , , if , , 0; and are the first parameters in the fault propagation chain and the real-time propagation chain after standardization, is the L1 norm of .

[0059] Wherein, the standardization is used to eliminate the dimension of abnormal interval and trend increment in the edge weight.

[0060] In this way, the probability of each fault type at any time is obtained, and the probability of any fault type at any time is greater than the probability threshold, indicating that the lithium battery power module corresponds to the corresponding fault type, and a warning is issued. The threshold is 0.5.

[0061] According to the second aspect of the application, the application also provides a lithium battery power module operation monitoring system. Figure 3 is a structural block diagram of a lithium battery power module operation monitoring system according to an embodiment of the application. As Figure 3As shown, the system 50 comprises a processor and a memory having stored thereon computer program instructions which, when executed by the processor, implement a lithium battery power module operation monitoring method according to the first aspect of the present application. The system further comprises a communication bus and a communication interface and other components well known to those skilled in the art, the arrangement and function of which are known in the art and thus will not be described here in detail.

[0062] It should be noted that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all fall within the scope of protection of the present application.

Claims

1. A lithium battery power module operation monitoring method, characterized in that: The monitoring method comprises: constructing a fault propagation chain for each fault type, including: performing anomaly detection on the time series of each parameter in the historical data of any fault type: dividing the time series of any parameter into multiple subsequences according to a preset window, obtaining the abnormal value of each subsequence by using the LOF algorithm, and the abnormal value of each subsequence constitutes the abnormal value sequence of the parameter, thereby obtaining the abnormal value sequence of each parameter; calculating the abnormal rising trend at each moment in the abnormal value sequence; and calculating the abnormal rising trend at each moment in the abnormal value sequence. Abnormal upward trend for: ; For the moment Neighborhood interval of ; and is a set of value pairs, representing the time In the neighborhood interval of and time Parameter value of Indicates the acquisition time The neighborhood interval of All value pairs of , is the mean value function, Take the value as 5; connect each parameter in ascending order of abnormal moment, where the abnormal moment is the moment when the abnormal rising trend first exceeds the threshold; use the abnormal interval and trend increment as the edge weight of each parameter other than the first parameter to obtain the fault propagation chain of the fault type; the abnormal interval and trend increment are the interval between the abnormal moments of any parameter and the previous adjacent parameter, and the difference in the abnormal rising trend at the abnormal moment, respectively; The abnormal value sequence corresponding to the real-time data of each parameter is obtained, and the abnormal rising trend of each parameter is monitored in real time to build a real-time propagation chain. The real-time propagation chain is compared with the fault propagation chain of each fault type to obtain the probability of each fault type at any time.

2. The method for monitoring the operation of a lithium battery power module according to claim 1, wherein: The parameters include voltage, current, internal resistance, battery temperature and state of charge.

3. The method for monitoring the operation of a lithium battery power module according to claim 1, wherein: The construction of the fault propagation chain for each fault type also includes counting the number of occurrences of each parameter sequence in the fault propagation chains corresponding to multiple historical data, determining the parameter sequence with the largest number of occurrences, and then calculating the average value of the edge weights of each parameter in all fault propagation chains corresponding to the parameter sequence to obtain the final fault propagation chain for the fault type.

4. The method for monitoring the operation of a lithium battery power module according to claim 1, wherein: The real-time monitoring of abnormal rising trends of various parameters to build a real-time transmission chain includes: In response to an abnormal moment of any parameter at the current moment, a real-time propagation chain with the parameter as the starting end is constructed at the current moment; after the current moment, if any other parameter other than the parameter has an abnormal moment without exceeding the maximum waiting time, the other parameter is connected to the terminal end of the real-time propagation chain, and the edge weight of the other parameter is calculated until all parameters are traversed and the construction of the real-time propagation chain is completed.

5. The method for monitoring the operation of a lithium battery power module according to claim 1, wherein: time Fault type Probability The steps to obtain include: Response and timing There is no established real-time propagation chain, the fault type Probability is 0, otherwise, calculate the time Real-time propagation chains and fault types The similarity between fault propagation chains, taking the maximum similarity as the moment Fault type Probability .

6. A lithium battery power module operation monitoring method according to claim 5, characterized in that: Real-time communication chain and fault type Fault propagation chain The similarity between for: ; For real-time communication chain and fault propagation chains length; and Fault propagation chain and real-time communication chain Middle parameters, if , ,like , 0; and Fault propagation chain and real-time communication chain Middle The edge weight after parameter standardization, for The L1 norm of .

7. The method for monitoring the operation of a lithium battery power module according to claim 1, wherein: Before obtaining the abnormal value sequence of each parameter, the monitoring method further includes: performing normalization processing on the time series of each parameter.

8. A lithium battery power module operation monitoring system, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a lithium battery power module operation monitoring method according to any one of claims 1 to 7 is implemented.

Images