Fault early warning method and device based on power battery voltage consistency

Abstract

The invention provides a fault early warning method and device based on power battery voltage consistency. The method comprises the following steps: acquiring voltage time sequence data of each minimum parallel unit in a battery pack in real time; the voltage time sequence data comprises voltage values at all moments; determining a consistency index sequence based on the voltage time sequence data; inputting the consistency index sequence into a pre-trained unsupervised learning reference model to obtain a real-time deviation degree score; and if the real-time deviation degree score exceeds a dynamic threshold value, triggering first-level early warning, and generating a first-level early warning log. According to the method, early warning and accurate positioning of the power battery fault are realized, and the safety management level of the power battery of the electric vehicle is remarkably improved. According to the method, early warning and accurate positioning of the power battery fault are realized, and the safety management level of the power battery of the electric vehicle is remarkably improved.

Description

Technical Field

[0001] This application relates to the field of safety monitoring technology for power batteries in new energy vehicles, and more specifically, to a fault early warning method and device based on the consistency of power battery voltage. Background Technology

[0002] Safety monitoring of power batteries is one of the core technologies in the electric vehicle field. Existing technologies mostly rely on setting thresholds based on parameters such as the battery's total voltage, highest / lowest single-cell voltage, or temperature to determine faults. For example, patent CN119840422A discloses a battery three-level alarm identification method based on a bidirectional long short-time memory network (Bi-LSTM). This method determines whether the battery is in a three-level alarm state by analyzing the temporal characteristics of vehicle signal data 61 seconds before and after the alarm time. This represents a technological direction of using advanced algorithms to improve alarm accuracy. Furthermore, GB / T32960.3-2025, "Technical Specifications for Remote Service and Management Systems for Electric Vehicles Part 3: Communication Protocols and Data Formats," requires monitoring the voltage data of the battery's "smallest parallel unit," aiming to achieve more refined monitoring.

[0003] However, the methods of the aforementioned existing technologies are essentially state recognition rather than early warning. They can usually only trigger an alarm when irreversible changes have occurred inside the battery (such as internal short circuits causing temperature rise or sudden voltage changes). This loses the time window for early intervention and makes it difficult to accurately locate which smallest parallel unit in the battery pack is abnormal, increasing the difficulty of troubleshooting and maintenance costs. Summary of the Invention

[0004] The purpose of this application is to provide a fault early warning method and device based on the consistency of power battery voltage, so as to realize early warning and accurate location of power battery faults, and significantly improve the safety management level of power battery of electric vehicles.

[0005] Firstly, a fault early warning method based on the voltage consistency of a power battery is provided, applied to an in-vehicle T-Box. This method may include: real-time acquisition of voltage time-series data of each smallest parallel unit within the battery pack; the voltage time-series data includes voltage values ​​at various times; determining a consistency index sequence based on the voltage time-series data; the consistency index sequence includes a final consistency index reflecting the voltage dispersion of all smallest parallel units at various times within a first time period; inputting the consistency index sequence into a pre-trained unsupervised learning benchmark model to obtain a real-time deviation score, wherein the benchmark model is trained based on historical final consistency indices during battery health periods; if the real-time deviation score exceeds a dynamic threshold, a first-level early warning is triggered, and a first-level early warning log is generated.

[0006] In one possible implementation, determining a consistency index sequence based on the voltage time series data includes: obtaining the average voltage and corresponding voltage standard deviation of any smallest parallel unit at each time point; taking any time point as the current time point, determining the average voltage standard deviation of the current time point and multiple consecutive time points before the current time point as the final consistency index; and determining the final consistency index of each time point within the first time period as a consistency index sequence.

[0007] In one possible implementation, the configuration method of the dynamic threshold includes: uploading multiple real-time deviation scores obtained within the fourth time period to the cloud platform, so that the cloud platform calculates a specified quantile of the real-time deviation score, and uses a weighted fusion algorithm to process the specified quantile and historical dynamic thresholds to obtain a new dynamic threshold; and obtaining the new dynamic threshold sent by the cloud platform.

[0008] In one possible implementation, the method further includes: after triggering the first-level warning, extracting the trigger time of the first-level warning and the warning time period including the trigger time; for each minimum parallel unit, determining at least one suspected abnormal unit based on the deviation between the voltage value of each minimum parallel unit at each moment in the warning time period and the average voltage at the corresponding moment; if the voltage deviation change of any suspected abnormal unit in the second time period meets the preset abnormal change condition, then confirming that the suspected abnormal unit is an abnormal unit; triggering a second-level warning and reporting the second-level warning information.

[0009] In one possible implementation, at least one suspected abnormal unit is determined based on the deviation between the voltage value of each minimum parallel unit at each moment during the warning period and the average voltage at the corresponding moment. This includes: calculating the average absolute deviation between the voltage value of each minimum parallel unit at each moment during the warning period and the average voltage at the corresponding moment; sorting the average absolute deviations of each minimum parallel unit in descending order, and selecting the minimum minimum parallel unit with the largest average absolute deviation as at least one suspected abnormal unit.

[0010] In one possible implementation, the preset abnormal change condition is that the voltage deviation change in the second time period is continuously higher than n times its average deviation value in the past third time period, where the third time period is longer than the second time period and n is an integer greater than 1.

[0011] In one possible implementation, the secondary warning information includes a unique identifier for the vehicle, a timestamp, a warning level, an anomaly score, and a list of anomalous units, the list of anomalous units containing the numbers of the anomalous units and their average voltage deviation values.

[0012] Secondly, a fault warning device based on the consistency of power battery voltage is provided, applied to an in-vehicle T-Box, which may include: The acquisition unit is used to acquire the voltage timing data of each smallest parallel unit in the battery pack in real time; the voltage timing data includes the voltage value at each moment. A determining unit is used to determine a consistency index sequence based on the voltage time series data; the consistency index sequence includes a final consistency index that reflects the voltage dispersion of all minimum parallel units at each moment within a first time period; The acquisition unit is used to input the consistency index sequence into a pre-trained unsupervised learning benchmark model to obtain a real-time deviation score, wherein the benchmark model is trained based on the historical final consistency index during the battery health period. The triggering unit is used to trigger a level one warning if the real-time deviation score exceeds the dynamic threshold. The generation unit is used to generate first-level early warning logs.

[0013] Thirdly, an electronic device is provided, which includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a program stored in memory, it implements any of the steps described in the first aspect above.

[0014] Fourthly, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when executed by a processor, the computer program implements the steps of any of the methods described in the first aspect above.

[0015] The fault early warning method and device based on power battery voltage consistency provided in this application involves real-time acquisition of voltage time-series data of each smallest parallel unit within the battery pack; the voltage time-series data includes the voltage value at each moment; based on the voltage time-series data, a consistency index sequence is determined; the consistency index sequence includes a final consistency index reflecting the voltage dispersion of all smallest parallel units at each moment within a first time period; the consistency index sequence is input into a pre-trained unsupervised learning benchmark model to obtain a real-time deviation score, the benchmark model being trained based on historical final consistency indices during battery health periods; if the real-time deviation score exceeds a dynamic threshold, a first-level early warning is triggered, and a first-level early warning log is generated. This method achieves early warning and accurate location of power battery faults, significantly improving the safety management level of electric vehicle power batteries. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the structure of a fault early warning system provided in an embodiment of this application; Figure 2 A flowchart illustrating a fault early warning method based on the consistency of power battery voltage provided in this application embodiment; Figure 3 A schematic diagram of a fault early warning device based on the consistency of power battery voltage provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application. Unless otherwise defined, the technical or scientific terms used in this application should have the ordinary meaning understood by those skilled in the art. The terms "first," "second," and similar terms used in this application do not indicate any order, quantity, or importance, but are only used to distinguish different components. The terms "comprising" or "including," etc., mean that the element or object preceding the word covers the element or object listed after the word and its equivalents, but does not exclude other elements or objects. The terms "connection," "coupled," or "linked," etc., are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect.

[0019] The fault early warning method based on power battery voltage consistency provided in this application embodiment can be applied to... Figure 1 In the fault early warning system shown, such as Figure 1 As shown, the fault warning system may include: a cloud platform, an in-vehicle T-Box, a battery management system (BMS), and a battery pack containing N minimum parallel units.

[0020] The cloud platform connects to the vehicle-mounted T-Box via a cellular network, the vehicle-mounted T-Box connects to the BMS via a CAN bus, and the BMS is electrically connected to the battery pack.

[0021] BMS is used to collect raw data such as voltage timing data, total current value, and temperature of each smallest parallel unit in the battery pack; the voltage timing data can include the voltage value at each moment. The vehicle-mounted T-Box is used to collect real-time voltage time-series data of each smallest parallel unit in the battery pack. The voltage time-series data includes the voltage value at each moment. Based on the voltage time-series data, a consistency index sequence is determined. The consistency index sequence includes the final consistency index that reflects the voltage dispersion of all smallest parallel units at each moment in the first time period. The consistency index sequence is input into a pre-trained unsupervised learning benchmark model to obtain a real-time deviation score. If the real-time deviation score exceeds the dynamic threshold, a first-level warning is triggered, and a first-level warning log is generated.

[0022] The cloud platform is used to train an unsupervised learning benchmark model based on historical eventual consistency metrics during battery health periods, and then send the model to the in-vehicle T-Box.

[0023] The system of this application can capture the earliest signs of faults such as slight increase in internal resistance and minor short circuit, and advance the alarm time from when the fault has already occurred to when the fault is brewing.

[0024] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0025] Figure 2 This is a flowchart illustrating a fault early warning method based on the consistency of power battery voltage, provided as an embodiment of this application. Figure 2 As shown, when applied to an in-vehicle T-Box, this method may include: Step S210: Real-time acquisition of voltage timing data of each smallest parallel unit in the battery pack.

[0026] Voltage timing data can include voltage values ​​at various times. Further data may include total current and battery pack temperature.

[0027] The onboard T-Box periodically receives data frames from the Battery Management System (BMS) via the Vehicle Controller Area Network (CAN) bus. Data acquisition follows the protocol specifications of standard GB / T32960.3-2025 "Technical Specifications for Remote Service and Management System of Electric Vehicles", which requires that the data frames contain the voltage values ​​of all the smallest parallel units in the battery pack.

[0028] Assuming the battery pack contains N minimum parallel units, then at each sampling time t, the T-Box acquires a voltage data vector: V(t) = [V1(t), V2(t), ..., V...]. N [(t)], where This represents the voltage value of the i-th smallest parallel unit at time t. The preferred sampling frequency is 1Hz, meaning that the voltage data of all units is acquired once per second. This ensures real-time performance without excessively consuming computational and communication resources. Simultaneously, the total current I(t) and battery pack temperature T(t) should be collected concurrently for subsequent auxiliary judgment and data validity verification.

[0029] Furthermore, in a real-world vehicle environment, data transmission may be affected by network fluctuations or electromagnetic interference, leading to outliers or missing values ​​in the raw data. Therefore, data preprocessing is necessary. 1) Outlier Removal: Outlier detection is performed using the 3σ criterion (also known as the Raida criterion). Specifically, for a specific minimum parallel unit i, a voltage data sequence within a continuous time window (e.g., 30 seconds) is selected. The mean of this sequence is calculated. and standard deviation For any data point within the window If it satisfies If the data point is an outlier, it will be removed from the data sequence.

[0030] 2) Missing value imputation: For outliers that have been removed or data points that were originally missing due to transmission interruptions, interpolation is used to fill in the missing values. Specifically, linear interpolation is used: assuming that cell i at time... Data Missing, but its previous moment and the next moment Data and If everything is normal, then use Interpolation is performed. If the missing value occurs at the beginning or end of the data sequence, it is filled with the nearest valid neighbor value, that is, the missing value is filled directly with the first or last valid data point.

[0031] The above preprocessing can effectively eliminate noise and interference, ensure the quality of the data on which subsequent analysis is based, and reduce the probability of false alarms from the source.

[0032] Step S220: Determine the consistency index sequence based on voltage timing data.

[0033] The core of this step is to transform the voltage vectors of the N units at each moment into a scalar index that can comprehensively reflect the consistency of the voltage inside the battery pack, and form a time sequence for model input.

[0034] The consistency index sequence can include the final consistency index reflecting the voltage dispersion of all minimum parallel units at each moment within the first time period (e.g., 6s).

[0035] (1) Obtain the mean voltage and corresponding standard deviation of any smallest parallel unit at each time point. The standard deviation of voltage is preferred as the core consistency indicator, as it can sensitively capture the dispersion of voltage across each unit. The calculation steps are as follows: For each sampling time t, calculate the average value of the N unit voltages: ; Calculate the voltage standard deviation σ(t) at time t: .

[0036] The unit of voltage standard deviation σ(t) is V. The smaller the value, the more concentrated the voltage of each unit and the better the consistency; the larger the value, the worse the consistency.

[0037] (2) Taking any time as the current time, the average of the voltage standard deviations of the current time and multiple consecutive times before the current time is determined as the final consistency index.

[0038] To eliminate the impact of instantaneous fluctuations, a sliding time window of length L (preferably L = 5 seconds) is introduced. The standard deviation sequence [σ(t-4), σ(t-3), ..., σ(t)] of the current time t and five points prior to it is calculated, and its average value is taken as the final consistency index at time t. : This method is equivalent to a low-pass filter, which can make the change curve of the consistency index smoother and make it easier to capture trend changes.

[0039] (3) The final consistency index at each time point in the first time period is determined as the consistency index sequence.

[0040] Step S230: Input the consistency index sequence into the pre-trained unsupervised learning benchmark model to obtain the real-time deviation score.

[0041] The baseline model is trained based on historical final consistency metrics during battery health periods. The model construction process can be completed on a cloud platform, with the aim of establishing a model that can represent the normal fluctuation range of consistency metrics when the battery is in a healthy state.

[0042] (1) Training data collection: Within the first 3 months or the first 5000 kilometers of driving after the vehicle leaves the factory (whichever comes first), the battery is usually in a healthy state (SOH>95%). The cloud platform continuously receives and stores pre-processed consistency index data uploaded from the T-Box. To form a training dataset , where M is the total amount of data.

[0043] (2) Algorithm selection and training: The Isolation Forest algorithm is preferred to build the baseline model. This algorithm is specifically used for anomaly detection, does not require labeled data (unsupervised learning), and has high computational efficiency, making it suitable for processing massive time series data.

[0044] (3) Brief description of the algorithm principle: Isolation Forest "isolates" data points by randomly selecting features and split values. Outliers are more likely to be isolated because they are significantly different from normal points (i.e., they require fewer splits). The number of splits required is the measure of deviation.

[0045] Training process: Cloud platform As input, the IsolationForest class from a machine learning library (such as Scikit-learn) is called to train the model. Key parameter settings are as follows: n_estimators (number of trees): Set to 100. Increasing the number of trees can make the model more stable, but it will increase the computational cost. 100 is an empirically balanced value.

[0046] Contamination (expected proportion of outliers in the dataset): Set to "auto" to let the algorithm determine it automatically.

[0047] max_samples (number of samples used per tree): set to "auto", which is 256.

[0048] After training is complete, the baseline model IF_model is obtained.

[0049] The training parameters for the Isolation Forest algorithm include: the number of trees is set to one hundred, the outlier expectation ratio is set to automatic detection, and the number of samples used per tree is set to automatic or fixed.

[0050] This step does not establish a fixed threshold, but rather a dynamic benchmark based on the battery's "health memory." It gives the system adaptive capabilities, enabling it to understand what constitutes "normal" fluctuations in the current battery pack, providing an intelligent and personalized benchmark for subsequent anomaly detection.

[0051] Furthermore, the T-Box will calculate the consistency index value in real time. The input is fed into the stored Isolation Forest model IF_model. The model outputs a real-time deviation score. .

[0052] Step S240: Trigger a Level 1 warning based on the real-time deviation score.

[0053] Compare the real-time deviation score with the dynamic threshold; If the real-time deviation score exceeds the dynamic threshold, a Level 1 warning is triggered and a Level 1 warning log is generated.

[0054] If the real-time deviation score does not exceed the dynamic threshold, monitoring needs to continue.

[0055] Furthermore, after a Level 1 warning is triggered, the trigger time of the Level 1 warning is extracted. and the warning period including the trigger time, i.e., the trigger time. One time window before and after (preferably with a window length of 10 seconds); For each minimum parallel unit, at least one suspected abnormal unit is identified based on the deviation between the voltage value of each minimum parallel unit at each moment during the warning period and the average voltage at the corresponding moment; specifically: calculate the average of the absolute deviations between the voltage value of each minimum parallel unit at each moment during the warning period and the average voltage at the corresponding moment. ,in, This is the warning period. It intuitively reflects the degree to which the i-th smallest parallel unit deviates from the overall average level during the warning period.

[0056] Then, the average absolute deviations of each smallest parallel unit are sorted in descending order, and the smallest parallel unit with the largest average absolute deviation is selected as at least one suspected abnormal unit.

[0057] Furthermore, determine the sustainability of the trend: If the voltage deviation change of any suspected abnormal unit in the second time period meets a preset abnormality condition, then the suspected abnormal unit is confirmed as an abnormal unit. The preset abnormality condition is that the voltage deviation change in the second time period is continuously higher than n times its average deviation value in the past third time period (e.g., 24 hours), where the third time period is longer than the second time period, and n is an integer greater than 1. For example, the system continues to monitor the voltage deviation change of this at least one suspected abnormal unit in a subsequent period (preferably 60 seconds). If it is found that the voltage deviation value D_a of at least one unit in the list (e.g., ID_a) is continuously higher than twice its average deviation value in the past 24 hours, then the behavior of that unit is confirmed as abnormal.

[0058] A Level 2 warning is triggered, and Level 2 warning information is reported. The Level 2 warning information may include the vehicle's unique identifier, timestamp, warning level, anomaly score, and a list of anomalous units, wherein the list of anomalous units includes the anomalous unit's number and average voltage deviation value.

[0059] The above implementation method achieves accurate fault location. By analyzing the data at the smallest parallel unit level and designing a sorting and filtering algorithm, the abnormal unit number is directly output, thereby improving operation and maintenance efficiency. Furthermore, after receiving a Level 2 warning, the cloud platform can highlight the vehicle's alarm status on the monitoring center's large screen and automatically generate a repair work order containing the specific fault unit number, greatly improving the efficiency and accuracy of after-sales maintenance.

[0060] In some embodiments, the threshold Threshold_current is not constant to accommodate the slow changes in battery performance as it ages. Therefore, the T-Box will update all recent (e.g., the past day) data weekly. The scores are uploaded to the cloud platform. The cloud platform recalculates the quantiles of these new scores and combines them with historical data to generate an updated threshold, Threshold_updated, which is then sent to the vehicle. This allows the early warning system to keep pace with the times and remain effective.

[0061] Specifically, the configuration method of dynamic threshold may include: uploading multiple real-time deviation scores obtained in the fourth time period (the past day) to the cloud platform so that the cloud platform can calculate the specified quantile of the real-time deviation score, and using a weighted fusion algorithm to process the specified quantile and historical dynamic threshold to obtain a new dynamic threshold; and obtaining the new dynamic threshold sent by the cloud platform.

[0062] In one specific embodiment, the onboard T-Box periodically (e.g., weekly) processes all real-time deviation scores calculated recently (e.g., within the past day). Uploaded to the cloud platform. After receiving this new data, the cloud platform first calculates this batch of data. The specified quantile of the score (e.g., the 95th quantile). Then, a weighted fusion algorithm is used to fuse this newly calculated quantile with the currently used historical dynamic threshold, for example: (Where α is a weighting factor, for example, 0.7), thus generating an updated dynamic threshold, Threshold_updated. The cloud platform encrypts and distributes this new threshold to the in-vehicle T-Box. The T-Box updates its local Threshold_current with the new threshold.

[0063] The fault early warning method based on power battery voltage consistency provided in this application adopts a complete set of technical solutions, including data preprocessing, consistency index sequence construction, unsupervised learning anomaly detection, dynamic threshold early warning, and precise unit positioning. It effectively utilizes the minimum parallel unit level data required by national standards, realizes early warning and precise positioning of power battery faults, and significantly improves the safety management level of electric vehicle power batteries.

[0064] Corresponding to the above method, this application also provides a fault early warning device based on the consistency of power battery voltage, such as... Figure 3 As shown, the device includes: The acquisition unit 310 is used to acquire the voltage timing data of each smallest parallel unit in the battery pack in real time; the voltage timing data includes the voltage value at each moment. The determining unit 320 is used to determine a consistency index sequence based on the voltage time series data; the consistency index sequence includes a final consistency index that reflects the voltage dispersion of all minimum parallel units at each moment within a first time period; The acquisition unit 330 is used to input the consistency index sequence into a pre-trained unsupervised learning benchmark model to obtain a real-time deviation score, wherein the benchmark model is trained based on the historical final consistency index during the battery health period. Triggering unit 340 is used to trigger a first-level warning if the real-time deviation score exceeds the dynamic threshold; Generation unit 350 is used to generate first-level early warning logs.

[0065] The functions of each functional unit of the fault early warning device based on power battery voltage consistency provided in the above embodiments of this application can be implemented through the above methods and steps. Therefore, the specific working process and beneficial effects of each unit in the fault early warning device based on power battery voltage consistency provided in the embodiments of this application will not be repeated here.

[0066] This application also provides an electronic device, such as... Figure 4 As shown, it includes a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440.

[0067] Memory 430 is used to store computer programs; When the processor 410 executes the program stored in the memory 430, it performs the following steps: Real-time acquisition of voltage timing data of each smallest parallel unit in the battery pack; the voltage timing data includes the voltage value at each moment; Based on the voltage time series data, a consistency index sequence is determined; the consistency index sequence includes a final consistency index that reflects the voltage dispersion of all minimum parallel units at each moment within the first time period; The consistency index sequence is input into a pre-trained unsupervised learning benchmark model to obtain a real-time deviation score, wherein the benchmark model is trained based on the historical final consistency index during the battery health period. If the real-time deviation score exceeds the dynamic threshold, a Level 1 warning is triggered, and a Level 1 warning log is generated.

[0068] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0069] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0070] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0071] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0072] The implementation methods and beneficial effects of the various components of the electronic device in the above embodiments for solving the problem can be found in [reference needed]. Figure 2The steps in the illustrated embodiments are used to implement the electronic device. Therefore, the specific working process and beneficial effects of the electronic device provided in this application will not be repeated here.

[0073] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores instructions that, when executed on a computer, cause the computer to perform any of the fault warning methods based on power battery voltage consistency described in the above embodiments.

[0074] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the fault warning methods based on power battery voltage consistency described in the above embodiments.

[0075] Those skilled in the art will understand that the embodiments in this application can be provided as methods, systems, or computer program products. Therefore, the embodiments in this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments in this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0076] This application describes embodiments of methods, apparatus (systems), and computer program products according to embodiments of this application with reference to flowchart illustrations and / or block diagrams. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0077] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0078] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0079] Although preferred embodiments have been described in this application, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of this application.

[0080] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the claims in this application and their equivalents, then this application also intends to include these modifications and variations.

Claims

1. A fault early warning method based on the consistency of power battery voltage, characterized in that, Applied to in-vehicle T-Boxes, the method includes: Real-time acquisition of voltage timing data of each smallest parallel unit in the battery pack; the voltage timing data includes the voltage value at each moment; Based on the voltage time series data, a consistency index sequence is determined; the consistency index sequence includes a final consistency index that reflects the voltage dispersion of all minimum parallel units at each moment within the first time period; The consistency index sequence is input into a pre-trained unsupervised learning benchmark model to obtain a real-time deviation score, wherein the benchmark model is trained based on the historical final consistency index during the battery health period. If the real-time deviation score exceeds the dynamic threshold, a Level 1 warning is triggered, and a Level 1 warning log is generated.

2. The method as described in claim 1, characterized in that, Based on the voltage time series data, a consistency index sequence is determined, including: Obtain the mean voltage and the corresponding standard deviation of any smallest parallel unit at each time point; Taking any time point as the current time point, the average of the voltage standard deviations of the current time point and multiple consecutive time points preceding the current time point is determined as the final consistency index; The final consistency index at each time point within the first time period is determined as the consistency index sequence.

3. The method as described in claim 1, characterized in that, The configuration methods for the dynamic threshold include: Multiple real-time deviation scores obtained in the fourth time period are uploaded to the cloud platform so that the cloud platform can calculate the specified quantile of the real-time deviation score and use a weighted fusion algorithm to process the specified quantile and the historical dynamic threshold to obtain a new dynamic threshold. Get the new dynamic threshold sent by the cloud platform.

4. The method as described in claim 1, characterized in that, The method further includes: After a Level 1 warning is triggered, the triggering time of the Level 1 warning and the warning period including the triggering time are extracted; For each minimum parallel unit, at least one suspected abnormal unit is identified based on the deviation between the voltage value of each minimum parallel unit at each moment during the warning period and the average voltage at the corresponding moment. If the voltage deviation change of any suspected abnormal unit in the second time period meets the preset abnormal change condition, then the suspected abnormal unit is confirmed as an abnormal unit. The Level II warning was triggered, and the Level II warning information was reported.

5. The method as described in claim 4, characterized in that, Based on the deviation between the voltage value of each minimum parallel unit at each moment during the warning period and the average voltage at the corresponding moment, at least one suspected abnormal unit is identified, including: Calculate the average absolute deviation between the voltage value of each smallest parallel unit at each moment during the warning period and the average voltage at the corresponding moment; After sorting the average absolute deviations of each smallest parallel unit in descending order, select the smallest parallel unit with the largest average absolute deviation as at least one suspected abnormal unit.

6. The method as described in claim 4, characterized in that, The preset abnormal change condition is that the voltage deviation change in the second time period is continuously higher than n times its average deviation value in the past third time period, where the third time period is longer than the second time period and n is an integer greater than 1.

7. The method as described in claim 4, characterized in that, The secondary warning information includes the vehicle's unique identifier, timestamp, warning level, anomaly score, and a list of anomalous units. The list of anomalous units includes the anomalous unit's number and average voltage deviation value.

8. A fault early warning device based on the consistency of power battery voltage, characterized in that, For use in vehicle-mounted T-Boxes, the device includes: The acquisition unit is used to acquire the voltage timing data of each smallest parallel unit in the battery pack in real time; the voltage timing data includes the voltage value at each moment. A determining unit is used to determine a consistency index sequence based on the voltage time series data; the consistency index sequence includes a final consistency index that reflects the voltage dispersion of all minimum parallel units at each moment within a first time period; The acquisition unit is used to input the consistency index sequence into a pre-trained unsupervised learning benchmark model to obtain a real-time deviation score, wherein the benchmark model is trained based on the historical final consistency index during the battery health period. The triggering unit is used to trigger a level one warning if the real-time deviation score exceeds the dynamic threshold. The generation unit is used to generate first-level early warning logs.

9. An electronic device, characterized in that, The electronic device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method of any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-7.

Images