Lithium ion battery fault automatic threshold optimization method and system and electronic equipment

Abstract

The invention provides a lithium ion battery fault automatic threshold optimization method and system and electronic equipment, and the method comprises the steps: extracting a plurality of battery time sequence data from a charging segment of a battery, including voltage, current, temperature and battery electric quantity; after battery time sequence data are coded and decoded, a deep learning model is used for training, so that each charging segment is coded into low-dimensional representation and then decoded into reconstructed data; calculating a reconstruction error of each charging segment based on the reconstruction data, and sorting the reconstruction errors to obtain sorted reconstruction errors; calculating the proportion of the fault vehicles based on the sorted reconstruction errors; an optimal reconstruction error threshold value is selected according to the proportion of the sorted reconstruction errors and the faulty vehicles, so that the proportion of the faulty vehicles is maximized; and updating the reconstruction error in real time when new data enter, and recalculating the reconstruction error according to the new data so as to update the optimal reconstruction error threshold in real time. The method solves the problem of threshold selection in lithium ion battery fault detection.

Description

Technical Field

[0001] The present invention relates to the technical field of battery fault detection, and in particular to a method, system and electronic equipment for automatically optimizing a lithium-ion battery fault threshold. Background Art

[0002] With the increasing popularity of electric vehicles, lithium-ion batteries (LiBs), as the core power system of electric vehicles, are widely used in various modes of transportation. LiBs have become the preferred battery type for electric vehicles due to their high energy density, long service life, and low self-discharge rate. However, LiBs may malfunction during use. These malfunctions often manifest as problems such as overcharging, over-discharging, inconsistent battery cell voltages, and excessively high or low temperatures. These malfunctions not only affect battery performance but also pose potential safety risks. Therefore, real-time monitoring and detection of the health status of LiBs, especially fault detection, has become a critical task in battery management systems (BMS).

[0003] In existing technologies, battery fault detection typically relies on fixed thresholds or manually defined rules. For example, multi-threshold methods for anomaly detection set multiple thresholds and utilize weighting strategies (such as linear weighting and aggregation) to determine the final anomaly detection result. Optimization algorithms adjust the weight of each threshold based on the importance of different features. However, this approach struggles to provide high-precision fault detection when faced with complex and diverse battery charging data.

[0004] In summary, the existing technology has the following shortcomings:

[0005] (1) Difficulty in threshold selection: In current fault detection methods, the selection of thresholds often relies on experience or static settings, which is difficult to automatically adjust according to the actual situation of the battery, thus affecting the accuracy of fault detection. For example, there may be dependencies between the thresholds of different features in the existing multi-threshold method based on anomaly detection, and these relationships need to be explicitly modeled in the model, otherwise it may lead to incorrect classification results.

[0006] (2) Poor adaptability to data changes: As battery charging data continues to increase, traditional methods are unable to dynamically adjust thresholds based on new data, resulting in outdated thresholds and inability to adapt to new data changes. For example, the existing multi-threshold method based on anomaly detection involves the weighting of multiple thresholds and multiple features. This method may increase computational complexity and require more computing resources. Summary of the Invention

[0007] In view of this, the present invention provides a lithium-ion battery fault automatic threshold optimization method, system and electronic equipment, aiming to solve the threshold selection problem in lithium-ion battery fault detection.

[0008] According to a first aspect of an embodiment of the present invention, a method for automatic threshold optimization of lithium-ion battery faults is provided, comprising: extracting multiple battery time series data from a charging segment of a battery, the battery time series data including voltage, current, temperature, and battery power; encoding and decoding the battery time series data, and then training using a deep learning model so that each charging segment is encoded into a low-dimensional representation and then decoded into reconstructed data; calculating a reconstruction error for each charging segment based on the reconstructed data and sorting the reconstruction errors to obtain a sorted reconstruction error; calculating the proportion of faulty vehicles based on the sorted reconstruction error; selecting an optimal reconstruction error threshold based on the sorted reconstruction error and the proportion of faulty vehicles to maximize the proportion of faulty vehicles; updating the reconstruction error in real time when new data enters, and recalculating the reconstruction error based on the new data to update the optimal reconstruction error threshold in real time.

[0009] Optionally, the deep learning model is an autoencoder or a variational autoencoder.

[0010] Optionally, the method selects an optimal reconstruction error threshold according to the sorted reconstruction error and the proportion of faulty vehicles to maximize the proportion of faulty vehicles, including: for a given current reconstruction error threshold τ, calculating the proportion P of faulty vehicles among all samples whose reconstruction errors are greater than the reconstruction error threshold. fault , the formula is as follows:

[0011]

[0012] Among them, Ⅱ(·) is the indicator function, Label i is the label of the i-th sample, Error i is the reconstruction error of the i-th sample, τ is the current reconstruction error threshold;

[0013] Select an optimal reconstruction error threshold τ oPt , which maximizes the proportion of faulty vehicles:

[0014] τ opt =argmaxP fault (τ)

[0015] Among them, τ opt is the optimal reconstruction error threshold.

[0016] Optionally, the updating formula of the optimal reconstruction error threshold is:

[0017]

[0018] Among them, τ new is the optimal reconstruction error threshold after real-time update, Error new is the new error value.

[0019] According to a second aspect of an embodiment of the present invention, a lithium-ion battery fault automatic threshold optimization system is provided, comprising: a data processing module for extracting multiple battery time series data from a battery charging segment, wherein the battery time series data includes voltage, current, temperature and battery power; a dynamic encoding and decoding module for encoding and decoding the battery time series data, and then training the battery time series data using a deep learning model so that each charging segment is encoded into a low-dimensional representation and then decoded into reconstructed data; a reconstruction error sorting module for calculating the reconstruction error of each charging segment based on the reconstructed data and sorting the reconstruction error to obtain the sorted reconstruction error; a reconstruction error calculation module for calculating the proportion of faulty vehicles based on the sorted reconstruction error; a threshold selection and fault detection module for selecting an optimal reconstruction error threshold based on the sorted reconstruction error and the proportion of faulty vehicles to maximize the proportion of faulty vehicles; a threshold updating module for updating the reconstruction error in real time when new data enters, and recalculating the reconstruction error based on the new data to update the optimal reconstruction error threshold in real time.

[0020] According to a third aspect of an embodiment of the present invention, there is provided an electronic device comprising a processor and a memory storing a program, wherein the program comprises instructions that, when executed by the processor, cause the processor to perform the steps of the method according to the first aspect.

[0021] According to a fourth aspect of an embodiment of the present invention, a computer storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the method of the first aspect is implemented.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] (1) The present invention avoids the dependency problem between feature thresholds by using a single threshold for fault detection, simplifies the model structure, and reduces unnecessary complexity.

[0024] (2) The present invention couples the principle of real-time inversion threshold of new errors, which can adapt to new data changes more efficiently, thereby improving the accuracy of fault detection and the adaptability of the system.

[0025] (3) Multi-threshold methods often require manual intervention to adjust the thresholds of various features, increasing the difficulty and workload of model debugging. The threshold optimization process of the present invention is completely automatic based on data. By maximizing the proportion of faulty vehicles, the optimal threshold is automatically selected, thus reducing the need for manual intervention and making the model more automated and intelligent. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0027] Figure 1 Flow chart of the steps of the method of the present invention.

[0028] Figure 2 4 is an overall flow chart of the method of the present invention.

[0029] Figure 3 Schematic diagram comparing the true and predicted values ​​of key features. DETAILED DESCRIPTION

[0030] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0031] See also Figure 1 The present invention provides a method for automatically optimizing a lithium-ion battery fault threshold, comprising:

[0032] Step S101: extracting a plurality of battery time series data from a battery charging segment, wherein the battery time series data includes voltage, current, temperature, and battery power;

[0033] Step S102: After encoding and decoding the battery time series data, a deep learning model is used for training so that each charging segment is encoded into a low-dimensional representation and then decoded into reconstructed data;

[0034] Step S103: calculating the reconstruction error of each charging segment based on the reconstructed data and sorting the reconstruction errors to obtain sorted reconstruction errors;

[0035] Step S104: Calculate the proportion of faulty vehicles based on the sorted reconstruction error;

[0036] Step S105: selecting an optimal reconstruction error threshold according to the sorted reconstruction error and the proportion of faulty vehicles to maximize the proportion of faulty vehicles;

[0037] Step S106: updating the reconstruction error in real time when new data enters, and recalculating the reconstruction error based on the new data to update the optimal reconstruction error threshold in real time.

[0038] Optionally, the deep learning model is an autoencoder or a variational autoencoder.

[0039] Optionally, the method selects an optimal reconstruction error threshold according to the sorted reconstruction error and the proportion of faulty vehicles to maximize the proportion of faulty vehicles, including: for a given current reconstruction error threshold τ, calculating the proportion P of faulty vehicles among all samples whose reconstruction errors are greater than the reconstruction error threshold. fault , the formula is as follows:

[0040]

[0041] Among them, Ⅱ(·) is the indicator function, Label i is the label of the i-th sample, Error i is the reconstruction error of the i-th sample, τ is the current reconstruction error threshold;

[0042] Select an optimal reconstruction error threshold τ opt , which maximizes the proportion of faulty vehicles:

[0043] τ opt =argmaxP fault (τ)

[0044] Among them, τ opt is the optimal reconstruction error threshold.

[0045] Optionally, the updating formula of the optimal reconstruction error threshold is:

[0046]

[0047] Among them, τ new is the optimal reconstruction error threshold after real-time update, Error new is the new error value.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] (1) The present invention avoids the dependency problem between feature thresholds by using a single threshold for fault detection, simplifies the model structure, and reduces unnecessary complexity.

[0050] (2) The present invention couples the principle of real-time inversion threshold of new errors, which can adapt to new data changes more efficiently, thereby improving the accuracy of fault detection and the adaptability of the system.

[0051] (3) Multi-threshold methods often require manual intervention to adjust the thresholds of various features, increasing the difficulty and workload of model debugging. The threshold optimization process of the present invention is completely automatic based on data. By maximizing the proportion of faulty vehicles, the optimal threshold is automatically selected, thus reducing the need for manual intervention and making the model more automated and intelligent.

[0052] Specifically, the solution of the present invention is further described according to the following examples:

[0053] This invention proposes an automatic threshold optimization method for lithium-ion battery faults. This method primarily optimizes fault detection accuracy and adapts to battery state changes by dynamically adjusting the reconstruction error threshold, overcoming the shortcomings of multi-threshold methods based on anomaly detection. Unlike multi-threshold methods, which require weighting thresholds for multiple features and exhibit high computational complexity, this invention achieves efficient and accurate fault detection through a dynamic optimization mechanism for a single threshold, combined with the principles of ranking reconstruction errors and maximizing the proportion of faulty vehicles. This method avoids dependencies between feature thresholds and eliminates the need for explicit modeling of multiple threshold relationships, thereby simplifying the model structure, reducing the computational burden, and improving the accuracy and adaptability of detection results.

[0054] See also Figure 2 , a lithium-ion battery fault automatic threshold optimization method of the present invention includes:

[0055] 1. Data processing and charging segment selection

[0056] This method first extracts multiple time-series data from the battery charging process, including important characteristics such as voltage, current, temperature, and battery charge (SOC). This data is segmented chronologically to analyze the battery's performance over different time periods.

[0057] 2. Dynamic encoding and decoding

[0058] Battery data is encoded and decoded, using deep learning models (such as autoencoders and variational autoencoders) for training. Each charging segment is encoded into a low-dimensional representation and then decoded into reconstructed data for calculating the reconstruction error.

[0059] 3. Reconstruction Error Calculation and Evaluation

[0060] The reconstruction error is calculated for each charging segment. The reconstruction error represents the difference between the original data and the model-reconstructed data, reflecting the degree of abnormality of the battery state.

[0061] In the evaluation phase, the reconstruction errors are first ranked and the proportion of faulty vehicles is calculated based on these errors.

[0062] 4. Threshold Optimization and Fault Detection

[0063] 1) For all samples {X1,X2,…,X n}, according to the calculated reconstruction error {Error1, Error2,…, Error n}, sort in descending order.

[0064] 2) For a given current reconstruction error threshold τ, calculate the proportion of faulty vehicles P among all samples whose reconstruction error is greater than the reconstruction error threshold fault , the formula is as follows:

[0065]

[0066] II(·) is an indicator function, which takes the value 1 if the condition is met and 0 otherwise;

[0067] Label i is the label of the i-th sample, 1 indicates fault and 0 indicates normal;

[0068] Error i is the reconstruction error of the i-th sample;

[0069] τ is the current reconstruction error threshold.

[0070] 3) Optimizing the abnormal distribution ratio, the present invention selects an optimal reconstruction error threshold τ opt , which maximizes the proportion of faulty vehicles:

[0071] τ opt =argmaxP fault (τ)

[0072] This step ensures the accuracy of fault detection and selects an optimal threshold to maximize the recognition rate of faulty vehicles.

[0073] 4) Couple the new error to the real-time inversion of the current reconstruction error threshold τ. When new data enters, the reconstruction error will be updated in real time. The model recalculates the reconstruction error based on the new data and dynamically adjusts the current reconstruction error threshold τ. The optimal reconstruction error threshold τ after real-time update new Perform inversion based on the latest reconstruction error distribution. Assume that the new error value is Error new , then the threshold update formula is:

[0074]

[0075] Among them, τ new is the optimal reconstruction error threshold after real-time update, Error new is the new error value.

[0076] See also Figure 3The feasibility of the automatic threshold optimization method for lithium-ion battery faults has been verified experimentally. Based on the sensitivity analysis of the features and the attention paid to key parameters in the electrochemical field, the present invention finally selected four important features for monitoring: SOC (state of charge), voltage difference, maximum temperature, and information entropy. These features have important physical significance in the battery system and can better reflect the working state and health of the system. By monitoring the error changes of these features, the model can gradually learn the rules in the data during the training process and detect potential faults in the battery in real time, providing effective support for early fault detection. From the training results of the model ( Figure 3 ), there was a large deviation between the model's predicted values ​​and the true values ​​in the initial stages (Epochs 2-3), indicating that the model had not yet fully learned the data characteristics and exhibited high error. Performance improved significantly after the seventh iteration (Epoch 7), demonstrating that the model was able to capture the key features of the data and exhibited good fitting and generalization capabilities. Entering the later stages of training (Epoch 12), the model's performance further stabilized, with the error fluctuations gradually decreasing and the curve fit remaining at a high level, indicating that the model was close to convergence and the training process achieved the expected results.

[0077] In summary, the present invention utilizes a threshold update mechanism based on real-time data error inversion. During the battery charging process, as new data enters in real time, the model calculates the reconstruction error and inversely updates the threshold τ, ensuring that the fault detection system can dynamically adapt to changes in the battery state and external environment. This mechanism dynamically adjusts the threshold based on real-time changes in the reconstruction error and the proportion of faulty vehicles, thereby improving the detection system's adaptability and accuracy. This mechanism ensures that the system can still effectively detect battery faults even when the data environment changes, avoiding the lag associated with traditional fixed-threshold methods. Furthermore, the present invention optimizes the abnormal distribution ratio by maximizing the proportion of faulty vehicles among all detected vehicles to optimize the fault detection threshold. Specifically, based on the relationship between the reconstruction error ranking and the proportion of faulty vehicles, the threshold τ that maximizes the proportion of faulty vehicles is selected. This method improves the precision and accuracy of fault detection without increasing computational complexity. Through this optimization method, the system maximizes the detection rate of faulty vehicles while ensuring a low false alarm rate for healthy vehicles, thereby providing more reliable detection results.

[0078] An embodiment of the present invention further provides a lithium-ion battery fault automatic threshold optimization system, comprising:

[0079] A data processing module is used to extract a plurality of battery time series data from the battery charging segment, wherein the battery time series data includes voltage, current, temperature and battery power;

[0080] The dynamic encoding and decoding module is used to encode and decode the battery time series data, and then use the deep learning model for training so that each charging segment is encoded into a low-dimensional representation and then decoded into reconstructed data;

[0081] A reconstruction error sorting module is used to calculate the reconstruction error of each charging segment based on the reconstructed data and sort the reconstruction errors to obtain sorted reconstruction errors;

[0082] A reconstruction error calculation module is used to calculate the proportion of faulty vehicles based on the sorted reconstruction errors;

[0083] The threshold selection and fault detection module is used to select an optimal reconstruction error threshold based on the sorted reconstruction error and the proportion of faulty vehicles, so as to maximize the proportion of faulty vehicles;

[0084] The threshold updating module is used to update the reconstruction error in real time when new data enters, and recalculate the reconstruction error based on the new data to update the optimal reconstruction error threshold in real time.

[0085] It should be understood that the system of this embodiment is used to implement the corresponding methods in the aforementioned multiple method embodiments and has the beneficial effects of the corresponding method embodiments.

[0086] As another example, the present invention also provides an electronic device, which will now be described as an electronic device that can serve as a server or client of the present invention, which is an example of a hardware device that can be applied to various aspects of the present invention. The electronic device is intended to represent various forms of digital electronic computer equipment, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0087] The electronic device may include: a processor, a communication interface, a memory, and a communication bus.

[0088] The processor, communication interface and memory communicate with each other through a communication bus. The communication interface is used to communicate with other electronic devices or servers.

[0089] The processor is used to execute programs, and specifically can execute the relevant steps in the above method embodiments.

[0090] Specifically, the program may include program codes including computer operation instructions.

[0091] The processor may be a CPU, an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present invention. The one or more processors included in the smart device may be processors of the same type, such as one or more CPUs, or processors of different types, such as one or more CPUs and one or more ASICs.

[0092] The memory is used to store programs and may include high-speed RAM memory or non-volatile memory, such as at least one disk storage.

[0093] When executed by a processor, the program is used to enable an electronic device to execute a lithium-ion battery fault automatic threshold optimization method of the present invention.

[0094] In addition, the specific implementation of each step in the program can refer to the corresponding description of the corresponding steps and units in the above method embodiments, and will not be repeated here. Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working process of the above-described devices and modules can refer to the corresponding process description in the above method embodiments, and will not be repeated here.

[0095] An exemplary embodiment of the present invention further provides a computer storage medium storing a computer program, wherein when the computer program is executed by a processor, the methods of the various embodiments of the present invention are implemented. The corresponding process descriptions in the aforementioned method embodiments can be referred to and will not be repeated here.

[0096] The method according to the embodiment of the present invention described above can be implemented in hardware, firmware, or as software or computer code that can be stored in a recording medium (such as a CD ROM, RAM, floppy disk, hard disk or magneto-optical disk), or as computer code that is originally stored in a remote recording medium or a non-temporary machine-readable medium downloaded via a network and will be stored in a local recording medium, so that the method described herein can be stored in such software processing on a recording medium using a general-purpose computer, a dedicated processor or programmable or dedicated hardware (such as an ASIC or FPGA). It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component (e.g., RAM, ROM, flash memory, etc.) that can store or receive software or computer code, and when the software or computer code is accessed and executed by a computer, a processor or hardware, the method described herein is implemented. In addition, when a general-purpose computer accesses the code for implementing the method shown here, the execution of the code converts the general-purpose computer into a dedicated computer for executing the method shown here.

[0097] Thus far, specific embodiments of the present invention have been described. Other embodiments are within the scope of the appended claims. In some cases, the actions recited in the claims can be performed in a different order and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing may be advantageous.

[0098] It should be understood that although this specification is described according to various embodiments, not every embodiment contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

[0099] Finally, it should be noted that the above implementation methods are only used to illustrate the embodiments of the present invention, and are not limitations on the embodiments of the present invention. Ordinary technicians in the relevant technical field can make various changes and modifications without departing from the spirit and scope of the embodiments of the present invention. Therefore, all equivalent technical solutions also fall within the scope of the embodiments of the present invention, and the scope of patent protection of the embodiments of the present invention should be defined by the claims.

Claims

1. A lithium-ion battery fault automatic threshold optimization method, characterized in that: include: Extracting a plurality of battery time series data from a battery charging segment, wherein the battery time series data includes voltage, current, temperature, and battery power; After encoding and decoding the battery time series data, a deep learning model is used for training, so that each charging segment is encoded into a low-dimensional representation and then decoded into reconstructed data; Calculating a reconstruction error of each charging segment based on the reconstructed data and sorting the reconstruction errors to obtain sorted reconstruction errors; Calculate the proportion of faulty vehicles based on the reconstruction error after sorting; An optimal reconstruction error threshold is selected based on the sorted reconstruction error and the proportion of faulty vehicles to maximize the proportion of faulty vehicles. The reconstruction error is updated in real time when new data comes in, and the reconstruction error is recalculated based on the new data to update the optimal reconstruction error threshold in real time.

2. The method according to claim 1, characterized in that The deep learning model is an autoencoder or a variational autoencoder.

3. The method according to claim 1, characterized in that The step of selecting an optimal reconstruction error threshold based on the sorted reconstruction error and the proportion of faulty vehicles to maximize the proportion of faulty vehicles includes: For a given current reconstruction error threshold τ, calculate the proportion of faulty vehicles P among all samples whose reconstruction error is greater than the reconstruction error threshold fault , the formula is as follows: Among them, Ⅱ(·) is the indicator function, Label i is the label of the i-th sample, Error i is the reconstruction error of the i-th sample, τ is the current reconstruction error threshold; Select an optimal reconstruction error threshold τ oPt , which maximizes the proportion of faulty vehicles: t opt =argmaxP fault (t) Among them, τ opt is the optimal reconstruction error threshold.

4. The method according to claim 3, characterized in that The updating formula of the optimal reconstruction error threshold is: Among them, τ new is the optimal reconstruction error threshold after real-time update, Error new is the new error value.

5. A lithium-ion battery fault automatic threshold optimization system, characterized in that: include: A data processing module is used to extract a plurality of battery time series data from the battery charging segment, wherein the battery time series data includes voltage, current, temperature and battery power; The dynamic encoding and decoding module is used to encode and decode the battery time series data, and then use the deep learning model for training so that each charging segment is encoded into a low-dimensional representation and then decoded into reconstructed data; A reconstruction error sorting module is used to calculate the reconstruction error of each charging segment based on the reconstructed data and sort the reconstruction errors to obtain sorted reconstruction errors; A reconstruction error calculation module is used to calculate the proportion of faulty vehicles based on the sorted reconstruction errors; The threshold selection and fault detection module is used to select an optimal reconstruction error threshold based on the sorted reconstruction error and the proportion of faulty vehicles, so as to maximize the proportion of faulty vehicles; The threshold updating module is used to update the reconstruction error in real time when new data enters, and recalculate the reconstruction error based on the new data to update the optimal reconstruction error threshold in real time.

6. An electronic device, characterized in that: include: processor; Memory for storing programs; The program includes instructions, which, when executed by the processor, cause the processor to perform the steps of the method according to any one of claims 1 to 4.

7. A computer storage medium, characterized in that A computer program is stored thereon, and when the program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images