Lithium ion battery anomaly detection method based on stream-oriented computation

Abstract

The invention discloses a flow calculation-based real-time anomaly detection method and system in a lithium ion battery system, and the method comprises the steps: achieving the real-time processing of battery data through a flow calculation frame, extracting the multi-dimensional features, such as variance, Q quantile difference, progressive decrease range sum, Z score, etc. And in combination with a multi-level early warning mechanism (such as primary, intermediate and advanced), accurate identification and hierarchical response of the abnormity are realized. Experimental results show that the method can effectively detect the abnormity of the single battery, reduce decision delay and improve the safety and reliability of an energy storage system, and is suitable for real-time monitoring and management of a large-scale battery pack.

Description

Technical Field

[0001] The present invention relates to the technical field of lithium-ion battery safety monitoring, and in particular to a real-time anomaly detection method and system for lithium-ion battery systems based on stream computing, which is applicable to battery management systems (BMS) of energy storage systems, electric vehicles, and portable electronic devices. Background Art

[0002] In the field of lithium-ion battery safety, existing anomaly detection algorithms have significant limitations, especially in terms of real-time performance and efficiency, which struggle to meet the ever-increasing demands for rapid response capabilities in modern applications. Traditional methods, such as those based on signals, data-driven methods, and models, suffer from high processing latency, high computational resource requirements, and insufficient feature sensitivity, making them unable to effectively address the real-time, dynamic, and complex nature of battery data. For example, data-driven methods rely on sample distribution assumptions, model-based methods require high-precision models and are computationally expensive, and traditional offline batch processing leads to decision lags, making it difficult to adapt to the real-time anomaly detection requirements of battery management systems. Summary of the Invention

[0003] In view of this, the first aspect of the present invention provides a lithium-ion battery anomaly detection method based on streaming computing to solve the problems of poor real-time performance, low computing efficiency, and inaccurate feature extraction in the existing technology, and realize real-time detection, graded warning, and efficient processing of lithium-ion battery anomalies. The method comprises the following steps: Collect lithium-ion battery data; Setting a trigger condition, and collecting the voltage of a single battery according to the trigger condition; Performing stream feature extraction on the single cell voltage, wherein the features include S variance, Q quantile difference, decreasing range sum, and Z score; Set up a multi-level early warning mechanism and implement multi-level early warning judgments on abnormal features.

[0004] Furthermore, the collecting of lithium-ion battery data specifically includes: Providing a battery pack comprising a plurality of cells; Clarify the specific connection method and number of cells in the battery pack; Clarify the battery pack parameters.

[0005] Specifically, the trigger condition is: when the current I>0 and the total voltage U total Reach the preset voltage point U record .

[0006] The collecting of the single cell voltage according to the trigger condition specifically includes: Use voltage and current signals as trigger conditions, set when current I>0 and total voltage U total Reach the preset voltage point U recordWhen the voltage of the single battery in the battery pack is triggered, the collection of the voltage of the single battery in the battery pack is triggered; When the trigger condition is met, the voltage Ui of the i-th single cell in the battery pack is collected in real time.

[0007] Specifically, The variance of S is F Var It is used to calculate the degree of deviation of each single cell voltage from the average voltage. The formula is: ; in, N is the number of battery cells, is the average voltage; The Q quantile difference F Q The formula used to calculate the voltage difference at different quantiles is: ; ; The decreasing range and F CDR It is used to calculate the cumulative voltage difference between the first N / 2 cells and the last N / 2 cells after sorting the voltage in descending order. The formula is: ; The Z score F Z It is used to identify voltage outliers and quantify the degree of deviation of data points from the mean. The formula is: ; in, is the voltage standard deviation.

[0008] Furthermore, the method further comprises the following steps: All extracted features are integrated and the four features are collectively referred to as F X , the eigenvalue recorded and calculated at sampling point t is recorded as F X (t).

[0009] Furthermore, the multi-level warning mechanism is specifically set up including: Set multi-level warning limits; According to the multi-level warning limit values, multi-level warning judgments are implemented in sequence; An early warning is issued based on the early warning judgment result.

[0010] Further details include: According to different characteristics, three warning limits K1, K2, and K3 are set; Make a primary warning (A3) judgment. When a certain feature meets An A3 primary warning is issued, indicating that the battery pack may show initial signs of inconsistency and requires early investigation; Make intermediate warning (A2) judgment and set the counter T Record the number of times the characteristic rises continuously. hour , and when When the counter returns to zero, ;when When the A2 intermediate warning is issued, it indicates that the abnormality persists and requires further close attention; Perform advanced warning (A1) judgment, when the characteristics When the A1 high-level warning is issued, it indicates that the battery pack has a serious abnormality and immediate intervention measures are required.

[0011] Furthermore, it also includes: Generate abnormality reports in real time and push them to the battery management system.

[0012] In a second aspect, a real-time anomaly detection system based on stream computing in a lithium-ion battery system is provided, based on the method described in the first aspect. It includes: a data acquisition module, configured to collect battery voltage and current signals; A stream processing engine, configured to perform data cleaning, streaming feature extraction, and multi-level warning calculations; The early warning and control module is configured to generate early warning reports and link the BMS to execute response strategies.

[0013] Beneficial effects of the present invention: 1. Improved real-time performance: Based on a stream computing framework, data is processed as it is generated. This eliminates the latency of traditional batch processing and reduces response time to seconds, meeting the real-time requirements of battery safety monitoring.

[0014] 2. Improved detection accuracy: The system integrates multi-dimensional features (variance, quantile, decreasing range, and Z score) to comprehensively characterize battery anomaly characteristics. Combined with a multi-level early warning mechanism, it effectively distinguishes transient fluctuations from persistent anomalies and reduces false alarm rates.

[0015] 3. Computational efficiency optimization: A lightweight feature extraction algorithm is used to reduce dependence on BMS computing resources. It is suitable for embedded systems and supports distributed real-time monitoring of large-scale battery packs.

[0016] 4. Enhanced scalability: Threshold parameters (K1, K2) can be dynamically adjusted according to different battery types (such as LFP), capacities (such as 100Ah) and application scenarios, with strong adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention has the following accompanying drawings: Figure 1 Flowchart of the stream computing anomaly detection framework; Figure 2This is a diagram of the streaming feature extraction framework; Figure 3 This is a diagram of the multi-layer alarm anomaly detection framework; Figure 4 The voltage curves of typical battery packs are shown in Figure 1. (a) is the abnormal battery pack No. 1, and (b) is the normal battery pack No. 2. Figure 5 The total voltage and characteristic value curves of abnormal battery pack No. 1 show the changing trend of characteristic values ​​over time. (a) is the total voltage curve of battery module No. 1, and (b) is the FX characteristic curve of battery module No. 1. Figure 6 This is the multi-level warning triggering sequence diagram for abnormal battery pack No. 1, verifying the effectiveness of the warning mechanism. (a) F var (b) F Q (c) F CDR ; (d) F Z DETAILED DESCRIPTION

[0018] In order to make the objectives, technical solutions and advantages of this application more clear, this application is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0019] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present application should have the usual meanings understood by people with ordinary skills in the field to which this application belongs. The "first", "second" and similar words used in the embodiments of the present application do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0020] like Figure 1 As shown in the figure, the flow chart of the stream computing anomaly detection framework shows the real-time processing flow from real-time data collection to analysis. For a battery pack containing multiple cells, the voltage and current signals are used as the main collection conditions, and streaming feature extraction is performed on the cell voltage to extract four practical, convenient and less redundant features: variance (F Var ), Q quantile difference (F Q), decreasing range and (F CDR ) and Z scores (F Z ), perform anomaly analysis on multiple features, adopt a multi-level warning mechanism and output warning results. The specific implementation process is as follows: S1: Provide a battery pack including multiple cells. This embodiment takes a battery pack of a certain energy storage system with 10 groups of 1 in parallel and 15 in series as an example, and collects voltage / current data for 6 months for verification.

[0021] S11: Determine the battery pack structure and provide a battery pack consisting of multiple single cells connected in series or in parallel, including but not limited to the 1P15S battery structure, and clarify the specific connection method and number of cells in the battery pack.

[0022] S12: Clarify the battery pack parameters, including the capacity and type of the battery pack, such as 100Ah capacity, LFP battery system type, and voltage range of 2.5-3.65V, to provide a benchmark for subsequent data collection and analysis.

[0023] S2: Use voltage and current signals as the main conditions to trigger the real-time acquisition of monomer voltage.

[0024] S21: Take voltage and current signals as the main acquisition conditions, set the current I>0 and the total voltage U total Reach the preset voltage point U record When the voltage of the single battery in the battery pack is triggered, the collection of the voltage of the single battery in the battery pack is triggered.

[0025] S22: When the triggering condition of S21 is met, the voltage Ui of the i-th single cell in the battery pack is collected in real time to ensure that the collected data reflects the stable characteristics of the battery in the charging state and reduce the interference of other states such as discharge on the data. Figure 4 , is the voltage curve of each battery cell collected, V1~V15 represents the voltage of 15 battery cells. (such as Figure 1 , (a) is normal battery pack No. 1, (b) is abnormal battery pack No. 2. ) S3: Extract the stream characteristics of the single cell voltage. (e.g. Figure 2 ) S31: Calculate the variance (F) using formula (1) Var ) feature. This feature is used to measure the discreteness of single cell voltage and reflect the inconsistency of the battery pack.

[0026] ; in, N is the number of battery cells, is the average voltage.

[0027] S32: Calculate the Q quantile difference (F) using formulas (2 and 3)Q ), in this embodiment, the quantile position is determined to be 0.25. This feature is used to capture different characteristics of voltage distribution. The focus of feature extraction can be adjusted by modifying the Q value, while saving computing costs.

[0028] ; ; S33: After arranging the battery cell voltages in descending order, the decreasing range sum is calculated using formula (4). This feature reflects the unevenness of the voltage distribution. When the battery cell voltages are inconsistent, the value will increase.

[0029] ; S34: Calculate the Z score using formula (5). This feature is used to identify outliers in a single cell and quantify the degree of deviation of a data point from the mean.

[0030] ; S35: Integrate all extracted features and refer to the four features as F X , the eigenvalue recorded and calculated at sampling point t is recorded as F X (t). Figure 5 , which is the total voltage and characteristic value curve of abnormal battery pack No.1, showing the changing trend of characteristic values ​​over time.

[0031] S4: Perform anomaly analysis on multiple features and adopt a multi-level warning mechanism. (e.g. Figure 3 ) S41: Set multi-level warning thresholds according to The thresholds of warnings at all levels are set by using the principle and other methods. Var set up , for F Z set up As shown in Table 1, these thresholds can be adjusted according to actual conditions in engineering applications.

[0032] Table 1 Thresholds for each feature ; S42: Make a primary warning (A3) judgment. When a certain feature meets An A3 primary warning is issued at this time, indicating that the battery pack may show initial signs of inconsistency and requires early investigation.

[0033] S43: Perform intermediate warning (A2) judgment and set the counter T Record the number of times the characteristic rises continuously. hour , and when When the counter returns to zero, .when When the A2 intermediate warning is issued, it indicates that the abnormality persists and requires further close attention.

[0034] S44: Perform advanced warning (A1) judgment, when the characteristics When the A1 advanced warning is issued, it indicates that the battery pack has a serious abnormality and requires immediate intervention. Figure 6 , the monitoring of abnormal battery pack No.1 shows that F Z At 2500 minutes, it first exceeds K2, triggering S3 warning; F Var and F CDR If the temperature rises continuously for more than 4 intervals at 5000 minutes, the S2 warning will be triggered; at 9500 minutes, the F Var >F CDR >K1, triggering S1 warning, which is consistent with the actual abnormal development trend.

[0035] S45: Process the warning results. If a warning is issued, appropriate action is taken based on the warning level. If no warning conditions are met, the process returns to S2 to continue data collection and analysis. Among 10 sets of actual monitoring data, three abnormal battery packs were successfully identified, all characterized by gradual deviations in single-cell voltage, validating the effectiveness and robustness of the method.

[0036] Based on the same design concept, the present invention proposes a real-time anomaly detection system based on stream computing in lithium-ion battery systems. The system includes: a data acquisition module configured to collect battery voltage and current signals; a stream processing engine configured to perform data cleaning, streaming feature extraction, and multi-level warning calculations; and a warning and control module configured to generate warning reports and link the BMS to execute response strategies.

[0037] It should be noted that any process or method description in the embodiments may be understood as representing a module, fragment or portion of code comprising one or more executable instructions for implementing steps of a specific logical function or process, and that the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0038] It should be noted that the logic and / or steps described in the embodiments, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device, or apparatus (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, device, or apparatus), or in conjunction with such an instruction execution system, device, or apparatus. For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by an instruction execution system, device, or apparatus, or in conjunction with such an instruction execution system, device, or apparatus. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion having one or more wires (electronic device), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and a portable compact disc read-only memory (CDROM). Furthermore, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing it in another suitable manner if necessary, and then storing it in a computer memory.

[0039] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0040] Those skilled in the art will understand that all or part of the steps of implementing the above-mentioned embodiment method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one of the steps of the method embodiment or a combination thereof.

[0041] Furthermore, the functional modules in the embodiments of the present invention may be integrated into a single processing module, each module may exist physically separately, or two or more modules may be integrated into a single module. The integrated modules may be implemented in the form of hardware or software functional modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium.

[0042] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0043] The above embodiments provide a detailed description of the technical solutions of the present invention. Obviously, the present invention is not limited to the described embodiments. Based on the embodiments of the present invention, those skilled in the art may make various modifications. However, any modifications that are equivalent to or similar to the present invention fall within the scope of protection of the present invention.

[0044] The contents not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A lithium-ion battery abnormality detection method based on stream computing, characterized in that: The steps include: Collect lithium-ion battery data; Setting a trigger condition, and collecting the voltage of a single battery according to the trigger condition; Performing stream feature extraction on the single cell voltage, wherein the features include S variance, Q quantile difference, decreasing range sum, and Z score; Set up a multi-level early warning mechanism and implement multi-level early warning judgments on abnormal features.

2. The method according to claim 1, characterized in that The collecting of lithium-ion battery data specifically includes: Providing a battery pack comprising a plurality of cells; Clarify the specific connection method and number of cells in the battery pack; Clarify the battery pack parameters.

3. The method according to claim 1, characterized in that The trigger condition is: when the current I>0 and the total voltage U total Reach the preset voltage point U record .

4. The method according to claim 3, characterized in that The collecting of the single cell voltage according to the trigger condition specifically includes: Use voltage and current signals as trigger conditions, set when current I>0 and total voltage U total Reach the preset voltage point U record When the voltage of the single battery in the battery pack is triggered, the collection of the voltage of the single battery in the battery pack is triggered; When the trigger condition is met, the voltage Ui of the i-th single cell in the battery pack is collected in real time.

5. The method according to claim 1, characterized in that The variance of S is F Var It is used to calculate the degree of deviation of each single cell voltage from the average voltage. The formula is: ; in, N is the number of battery cells, is the average voltage; The Q quantile difference F Q The formula used to calculate the voltage difference at different quantiles is: ; ; The decreasing range and F CDR It is used to calculate the cumulative voltage difference between the first N / 2 cells and the last N / 2 cells after sorting the voltage in descending order. The formula is: ; The Z score F Z It is used to identify voltage outliers and quantify the degree of deviation of data points from the mean. The formula is: ; in, is the voltage standard deviation.

6. The method according to claim 5, characterized in that The following steps are also included: All extracted features are integrated and the four features are collectively referred to as F X , the eigenvalue recorded and calculated at sampling point t is recorded as F X (t).

7. The method according to claim 6, characterized in that The multi-level early warning mechanism is specifically set up as follows: Set multi-level warning limits; According to the multi-level warning limit values, multi-level warning judgments are implemented in sequence; An early warning is issued based on the early warning judgment result.

8. The method according to claim 7, characterized in that Further details include: According to different characteristics, three warning limits K1, K2, and K3 are set; Perform primary warning A3 judgment, when a certain feature meets An A3 primary warning is issued when the battery pack shows initial signs of inconsistency and requires early investigation; Perform intermediate warning A2 judgment and set the counter T Record the number of times the characteristic rises continuously. hour , and when When the counter returns to zero, ;when When the A2 intermediate warning is issued, it indicates that the abnormality persists and requires further close attention; Perform advanced warning A1 judgment, when the characteristics When the A1 high-level warning is issued, it indicates that the battery pack has a serious abnormality and immediate intervention measures are required.

9. The method according to claim 8, characterized in that Also includes: Generate abnormality reports in real time and push them to the battery management system.

10. A real-time anomaly detection system based on stream computing in a lithium-ion battery system, characterized in that: Based on the method according to any one of claims 1 to 9, It includes: a data acquisition module, configured to collect battery voltage and current signals; A stream processing engine, configured to perform data cleaning, streaming feature extraction, and multi-level warning calculations; The early warning and control module is configured to generate early warning reports and link the BMS to execute response strategies.

Images