A method, apparatus, device, medium, and product for determining a battery state

Abstract

The application relates to a battery state determination method, device, equipment, medium and product. The method comprises at least the following steps: collecting state parameters of a battery and environmental parameters in which the battery is located; determining state characteristics of the battery based on the state parameters and the environmental parameters; performing time-dimension abnormality analysis on the state characteristics through a first model to obtain a first result of the battery; the first result is used for representing whether the battery is abnormal in the time dimension; performing space-dimension abnormality analysis on the state characteristics through a second model to obtain a second result of the battery; the second result is used for representing whether the battery is abnormal in the space dimension; and determining a target result used for representing the state of the battery based on the first result and the second result. The determined battery state is more accurate and has higher robustness, and the safety of the battery use is improved.

Description

Technical Field

[0001] This application relates to the field of battery technology, and in particular to a method, apparatus, device, medium, and product for determining the state of a battery. Background Technology

[0002] With the rapid development of electric vehicle technology, batteries, as a core component of electric vehicles, directly affect the vehicle's performance, safety, and reliability. As the electric vehicle market expands, consumers' demands for battery performance and safety are also increasing. However, batteries may experience various abnormalities during use, such as overheating, overcharging, over-discharging, and increased internal resistance. These abnormalities not only affect battery lifespan but may also lead to safety accidents such as fires and explosions.

[0003] Typical battery monitoring methods rely on simple threshold detection, such as setting fixed thresholds for parameters like voltage, temperature, and current. Once these thresholds are exceeded, the system issues an alarm. However, methods that determine battery status through thresholds fail to identify the relationships between various parameters, resulting in low accuracy. Summary of the Invention

[0004] One of the purposes of this application is to provide a method, apparatus, device, medium, and product for determining battery status, which determines the battery status more accurately and robustly, thereby improving the safety of battery use.

[0005] To achieve the above objectives, the technical solution adopted in this application is as follows:

[0006] In a first aspect, this application provides a method for determining the state of a battery, the method comprising: collecting state parameters of the battery and environmental parameters of the battery; determining state characteristics of the battery based on the state parameters and environmental parameters; performing anomaly analysis on the state characteristics in the time dimension using a first model to obtain a first result of the battery; the first result being used to characterize whether the battery is abnormal in the time dimension; performing anomaly analysis on the state characteristics in the spatial dimension using a second model to obtain a second result of the battery; the second result being used to characterize whether the battery is abnormal in the spatial dimension; and determining a target result for characterizing the state of the battery based on the first result and the second result.

[0007] Based on the aforementioned technical means, on the one hand, regarding data, this scheme considers not only the influence of battery state parameters on battery state but also the influence of environmental parameters on battery state, improving the accuracy of battery state determination. On the other hand, when determining the target result of battery state, the data is processed through models, improving the robustness of battery state determination. Furthermore, this scheme can achieve anomaly analysis in the time dimension through a first model and analysis in the spatial dimension through a second model, and then combine the results of both to determine the target result, further improving the accuracy of the target result. Therefore, the battery state determined by this scheme is more accurate and more robust, improving the safety of battery use.

[0008] In one possible implementation, determining a target result for characterizing the state of the battery based on a first result and a second result includes: if the first result and the second result are the same, then determining the state of the battery as the first result; if the first result and the second result are different, then determining a target score based at least on a first score and a second score; determining a target result based on the target score; wherein the first score is the score output by a first model; and the second score is the score output by a second model.

[0009] Based on the above technical means, when determining the target result, it is determined based on the results of both. Only when the first result and the second result are the same, the target result is obtained directly. When the first result and the second result are different, the target result is not determined directly by using one result as the standard. Instead, the first result and the second result are integrated through a more granular scoring method to determine the final target result, which improves the accuracy of the target result.

[0010] In one possible implementation, the target score is determined based at least on the first score and the second score, including: determining the target score based on the first score and the second score; and determining the target score based on the first score, the second score, and the feature score.

[0011] Based on the above technical means, the target score is determined based on the first score and the second score. When the two scores are merged, it has the characteristics of simple implementation and good interpretability. The target score is determined based on the first score, the second score and the feature score. It not only merges the two scores, but also merges the feature score, which further improves the accuracy of the target score, the accuracy of the target result and the safety of the battery.

[0012] In one possible implementation, determining a target score based on a first score, a second score, and a feature score includes: determining a score for each state parameter and a score for each environmental parameter; determining a feature score based on the score for each state parameter, the weight of each state parameter, the score for each environmental parameter, and the weight of each environmental parameter; and determining the target score based on the feature score, the weight of the feature score, the first score, the weight of the first score, the second score, and the weight of the second score.

[0013] Based on the above technical means, when determining the target score, the scores of each state parameter and each environmental parameter are considered, which improves the accuracy of the target score. Furthermore, weights are assigned to each state parameter and each environmental parameter, and the target score can be adjusted by adjusting the weights, which further improves the accuracy and flexibility of the target score.

[0014] In one possible implementation, the method further includes: if the battery status in the target result is abnormal, then determining the target abnormality type and target abnormality level of the battery; and outputting the target abnormality type, target abnormality level, and warning information.

[0015] Based on the above technical means, it is also possible to output the target anomaly type, target anomaly level, and early warning information, thereby improving the richness of the output and the user experience.

[0016] In one possible implementation, outputting early warning information includes: when the target anomaly level is Level 1, outputting a first early warning message via visual cues; when the target anomaly level is Level 2, outputting the first early warning message via visual cues and a second early warning message via auditory cues; when the target anomaly level is Level 3, outputting the first early warning message via visual cues, outputting the second early warning message via auditory cues, and outputting a third early warning message to the client via a network; wherein the anomaly severity of Level 1, Level 2, and Level 3 increases sequentially.

[0017] Based on the aforementioned technical means, different output methods are configured for different anomaly levels. For low anomaly levels, output is only provided visually to reduce the impact of the output on other processing. For high anomaly levels, multiple output methods are used to determine the reachability of the output, thereby improving security. In short, this multi-level output method improves the output effectiveness.

[0018] In one possible implementation, determining the state characteristics of the battery based on state parameters and environmental parameters includes: determining the rate of change of each state parameter and the rate of change of each environmental parameter respectively; and determining the state characteristics of the battery based on the rate of change of each state parameter and the rate of change of each environmental parameter.

[0019] Based on the aforementioned technical methods, when determining the state characteristics of a battery, the characteristics are not directly determined based on parameters. Instead, the rate of change of each state parameter and each environmental parameter is determined separately. The battery's state characteristics are then determined based on these rates of change. Since the rate of change better reflects the battery's dynamic performance, the obtained state characteristics are more consistent with reality, improving the accuracy of the state characteristics, the accuracy of the target results, and the battery's safety.

[0020] In one possible implementation, before determining the battery state characteristics based on the rate of change of each state parameter and the rate of change of each environmental parameter, the method further includes: removing outliers from the rate of change of each state parameter based on a first method; the first method includes at least one of a moving average algorithm, an autocorrelation function, and a wavelet transform; and removing outliers from the rate of change of each environmental parameter based on a moving average algorithm.

[0021] Based on the above technical means, different methods for removing outliers are configured according to the characteristics of different parameters, which effectively improves the accuracy of outlier removal.

[0022] In one possible implementation, when the state parameters include battery temperature, voltage, current, internal resistance, and state of charge, and the environmental parameters include ambient temperature, the state parameters of the battery and the environmental parameters of the battery are collected, including: collecting the battery voltage and current based on a first frequency; and collecting the battery temperature, internal resistance, state of charge, and ambient temperature based on a second frequency.

[0023] Based on the aforementioned technical methods, different frequencies are used for data acquisition depending on the characteristics of different parameters. This ensures the effectiveness of data acquisition while being applicable to the specific features of each parameter. For example, voltage and current parameters change rapidly, so a higher acquisition frequency is needed. Temperature parameters, on the other hand, generally do not change abruptly, so a slightly lower acquisition frequency can be used. This improves the accuracy of parameter acquisition, the accuracy of target results, and battery safety.

[0024] In one possible implementation, acquiring battery voltage and current based on a first frequency includes: determining the battery's operating state; if the battery's operating state is a first load state, increasing the first frequency; acquiring battery voltage and current based on the increased first frequency; if the battery's operating state is a second load state, decreasing the first frequency, and acquiring battery voltage and current based on the decreased first frequency; wherein the load of the first load state is higher than the load of the second load state.

[0025] Based on the above technical means, the acquisition frequency can be adjusted according to the battery's working state, which further improves the data acquisition efficiency, the accuracy of the target results, and the safety of the battery.

[0026] Secondly, this application provides a battery state determination device, the device comprising:

[0027] The data acquisition unit is used to collect the battery's status parameters and the environmental parameters in which the battery is located;

[0028] The first determining unit is used to determine the state characteristics of the battery based on state parameters and environmental parameters;

[0029] The first analysis unit is used to perform anomaly analysis on the state characteristics in the time dimension using the first model to obtain the first result of the battery; the first result is used to characterize whether the battery is abnormal in the time dimension.

[0030] The second analysis unit is used to perform spatial dimension anomaly analysis on the state characteristics using the second model to obtain the second result of the battery; the second result is used to characterize whether the battery is abnormal in the spatial dimension.

[0031] The second determining unit is used to determine the target result for characterizing the state of the battery based on the first result and the second result.

[0032] Thirdly, this application also provides an electronic device, which includes a processor and a memory, wherein the memory stores a computer program or instructions, which, when executed by the processor, implement the method provided in the first aspect.

[0033] Fourthly, this application also provides a storage medium storing a computer program or instructions that, when executed by a processor, implement the method provided in the first aspect above.

[0034] Fifthly, this application also provides a computer program product, which includes a computer program or instructions that, when executed by a processor, implement the method provided in the first aspect.

[0035] It should be noted that the technical effects of the second to fifth aspects can be referred to in the detailed description of the first aspect above, and will not be repeated here. Attached Figure Description

[0036] Figure 1 A schematic diagram of a first optional process for determining battery state provided in an embodiment of this application;

[0037] Figure 2 A schematic diagram of a second optional method for determining battery state provided in an embodiment of this application;

[0038] Figure 3 A schematic diagram of a third optional method for determining battery state provided in an embodiment of this application;

[0039] Figure 4 A schematic diagram of a fourth optional method for determining battery state provided in an embodiment of this application;

[0040] Figure 5 A schematic diagram of a fifth optional method for determining battery state provided in an embodiment of this application;

[0041] Figure 6 A schematic diagram of a sixth optional method for determining battery state provided in an embodiment of this application;

[0042] Figure 7 A schematic diagram of a seventh optional method for determining battery state provided in an embodiment of this application;

[0043] Figure 8 A schematic diagram of an eighth optional method for determining battery state provided in an embodiment of this application;

[0044] Figure 9 A schematic diagram of a ninth optional method for determining battery state provided in an embodiment of this application;

[0045] Figure 10 A schematic diagram of an optional structure of the electric vehicle battery abnormality warning device provided in an embodiment of this application;

[0046] Figure 11 A schematic diagram of an optional feature extraction process provided in an embodiment of this application;

[0047] Figure 12 A schematic diagram of an optional model processing procedure provided in an embodiment of this application;

[0048] Figure 13 A schematic diagram of an optional anomaly detection process provided in an embodiment of this application;

[0049] Figure 14 This is a schematic diagram of an optional structure of the battery state determination device provided in an embodiment of this application. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the specific technical solutions of the application will be further described in detail below with reference to the accompanying drawings of the embodiments of this application. The following embodiments are used to illustrate this application, but are not intended to limit the scope of this application.

[0051] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0052] In the following description, the terms "first," "second," and "third" are used only to distinguish different objects and do not represent a specific order of objects, nor are they constituting a chronological order. It is understood that "first," "second," and "third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0054] This application provides methods, apparatus, devices, media, and products for determining battery state. The method for determining battery state is executed by a battery state determining device, which can be deployed in an electronic device. The following describes various embodiments of the methods, apparatus, devices, media, and products for determining battery state provided in this application.

[0055] Firstly, embodiments of this application provide a method for determining battery state. The method will be described below using an electronic device as an example.

[0056] refer to Figure 1 The process shown may include, but is not limited to, S101 to S105.

[0057] S101, The electronic device collects the battery's status parameters and the environmental parameters of the battery's environment.

[0058] The embodiments of this application do not limit the type of battery, which can be determined according to the actual situation. For example, the battery here may include, but is not limited to: lithium-ion battery, lead-acid battery, lithium polymer battery, lithium iron phosphate battery, and ternary lithium battery, etc.

[0059] State parameters are used to characterize the current state of the battery. This application does not limit the type of state parameters; they can be configured according to actual needs. These state parameters may include, but are not limited to, at least one of the following: battery temperature, battery voltage, battery current, battery internal resistance, and battery state of charge.

[0060] Environmental parameters are used to characterize the current environmental state of the battery. This application does not limit the type of environmental parameters and they can be configured according to actual needs. These environmental parameters may include, but are not limited to, at least one of the following: ambient temperature and ambient humidity.

[0061] S101 can be implemented as follows: The electronic device collects the battery's state parameters and the environmental parameters of the battery's environment through a relevant acquisition device. The collected state parameters and environmental parameters can also be stored.

[0062] The status and environmental parameters collected here include the timestamp of the collection time, which can be used for subsequent analysis of parameter changes.

[0063] S102. The electronic device determines the state characteristics of the battery based on state parameters and environmental parameters.

[0064] State features are used to characterize the state of a battery through features.

[0065] S102 can be implemented as follows: the electronic device processes the state parameters and environmental parameters through a feature extraction algorithm to obtain the state characteristics of the battery.

[0066] S103. The electronic device performs anomaly analysis on the state characteristics in the time dimension using the first model to obtain the first result of the battery.

[0067] The first result is used to characterize whether the battery is abnormal in the time dimension.

[0068] The first model is used to perform anomaly analysis on state characteristics over time to determine whether the battery is abnormal and obtain a first result. This application embodiment does not limit the type of the first model and can be configured according to actual needs.

[0069] For example, the first model can be a Long Short-Term Memory (LSTM) network model or a gated recurrent unit model.

[0070] The first model here is a pre-trained model used to obtain the first result based on the state features.

[0071] S103 can be implemented as follows: the electronic device inputs the state characteristics into the first model, and performs anomaly analysis on the state characteristics in the time dimension through the first model to obtain the first result of the battery.

[0072] The first result here can include whether the result is abnormal. It can also include the type of abnormality and the level of abnormality, etc.

[0073] S104. The electronic device performs spatial dimension anomaly analysis on the state characteristics using the second model to obtain the second result of the battery.

[0074] The second result is used to characterize whether the battery is abnormal in the spatial dimension.

[0075] The second model is used to perform anomaly analysis on state characteristics in the spatial dimension to determine whether the battery is abnormal, thus obtaining a second result. This application embodiment does not limit the type of the second model and can be configured according to actual needs.

[0076] For example, the second model could be a random forest or a decision tree, etc.

[0077] The second model here is a pre-trained model used to obtain the second result based on the state features.

[0078] S104 can be implemented as follows: the electronic device inputs the state features into the second model, and performs spatial dimension anomaly analysis on the state features through the second model to obtain the second result of the battery.

[0079] The second result here can include whether the result is abnormal. It can also include the type of abnormality and the level of abnormality, etc.

[0080] S105, the electronic device determines the target result for characterizing the state of the battery based on the first result and the second result.

[0081] The target result is used to characterize the state of the battery as determined in this embodiment.

[0082] The target results here can include whether the result is abnormal. It can also include the target anomaly type and the target anomaly level, etc.

[0083] S105 can be implemented as follows: The electronic device determines a target result for characterizing the state of the battery based on the first result and the second result. If the first result and the second result are the same, the target result is determined to be the same result. If the first result and the second result are different, the target result can be one of the two or the result fused together.

[0084] In this embodiment, the method includes: collecting battery state parameters and environmental parameters of the battery; determining battery state characteristics based on the state parameters and environmental parameters; performing time-dimensional anomaly analysis on the state characteristics using a first model to obtain a first result for the battery; the first result is used to characterize whether the battery is abnormal in the time dimension; performing spatial-dimensional anomaly analysis on the state characteristics using a second model to obtain a second result for the battery; the second result is used to characterize whether the battery is abnormal in the spatial dimension; and determining a target result to characterize the battery state based on the first and second results.

[0085] Based on the aforementioned technical means, on the one hand, regarding data, this scheme considers not only the influence of battery state parameters on battery state but also the influence of environmental parameters on battery state, improving the accuracy of battery state determination. On the other hand, when determining the target result of battery state, the data is processed through models, improving the robustness of battery state determination. Furthermore, this scheme can achieve anomaly analysis in the time dimension through a first model and analysis in the spatial dimension through a second model, and then combine the results of both to determine the target result, further improving the accuracy of the target result. Therefore, the battery state determined by this scheme is more accurate and more robust, improving the safety of battery use.

[0086] The process by which the electronic device in S105 determines the target result for characterizing the state of the battery based on the first result and the second result will be explained below.

[0087] refer to Figure 2 The process may include, but is not limited to, S1051 and S1052 described below.

[0088] S1051. If the first result and the second result are the same, the electronic device determines the state of the battery as the first result.

[0089] Alternatively, the state of the battery can be considered as the second result, meaning the target result is either the first result or the second result.

[0090] S1051 can be implemented as follows: the electronic device compares whether the first result and the second result are the same. If the first result and the second result are the same, the state of the battery is determined to be the first result.

[0091] S1052. If the first result and the second result are different, the electronic device shall determine the target score based at least on the first score and the second score; and determine the target result based on the target score.

[0092] The first score is the score output by the first model; the second score is the score output by the second model.

[0093] In one possible implementation, the first score can be the anomaly probability output by the first model, and the second score can be the anomaly voting ratio output by the second model.

[0094] In another possible implementation, the first score can be the score converted from the anomaly probability output by the first model, and the second score can be the score converted from the anomaly voting ratio output by the second model.

[0095] The conversion method here can be normalization, so as to convert the two ratings to the same rating range.

[0096] S1052 can be implemented as follows: if the first result and the second result are different, the electronic device determines the target score based on the first score and the second score; or it can combine other parameters to determine the target score; and then determine the target result based on the target score.

[0097] Based on the above technical means, when determining the target result, it is determined based on the results of both. Only when the first result and the second result are the same, the target result is obtained directly. When the first result and the second result are different, the target result is not determined directly by using one result as the standard. Instead, the first result and the second result are integrated through a more granular scoring method to determine the final target result, which improves the accuracy of the target result.

[0098] The process by which the electronic device in S1052 determines the target score based at least on the first score and the second score will be explained below.

[0099] refer to Figure 3 The process may include, but is not limited to, S301 and S302 described below.

[0100] S301. The electronic device determines the target score based on the first score and the second score.

[0101] In one possible implementation, S301 can be implemented as follows: the electronic device determines the average of the first score and the second score as the target score.

[0102] In another possible implementation, S301 can be implemented as follows: the electronic device can further configure different weights for the first score and the second score, and determine the target score by the weighted average of the first score and the second score. This application embodiment does not limit the configured values ​​of the weights of the first score and the second score, and can configure them according to actual needs or experience.

[0103] S302. The electronic device determines the target score based on the first score, the second score, and the feature score.

[0104] Feature scoring refers to the scoring obtained based on various environmental and state parameters.

[0105] In one possible implementation, feature scores are obtained based on state features.

[0106] In another possible implementation, feature scores are determined based on environmental and state parameters.

[0107] S302 can be implemented as follows: the electronic device first obtains a feature score based on state characteristics; or determines a feature score based on environmental parameters and state parameters, and then determines a target score based on the first score, the second score, and the feature score.

[0108] Based on the above technical means, the target score is determined based on the first score and the second score. When the two scores are merged, it has the characteristics of simple implementation and good interpretability. The target score is determined based on the first score, the second score and the feature score. It not only merges the two scores, but also merges the feature score, which further improves the accuracy of the target score, the accuracy of the target result and the safety of the battery.

[0109] The process by which the electronic device in S302 determines the target score based on the first score, the second score, and the feature score will be explained below.

[0110] refer to Figure 4 The process may include, but is not limited to, S401 to S403 described below.

[0111] S401. The electronic device determines the score for each state parameter and the score for each environmental parameter.

[0112] The scoring method for determining state parameters can be the same as or different from the scoring method for determining environmental parameters.

[0113] For example, the score for each state parameter and the score for each environmental parameter can be determined based on formula (1).

[0114] Formula (1);

[0115] In formula (1), Characterizing the eigenvalues ​​of each parameter x, This is the standard value for the parameter; the normal range for parameter x is [ ].

[0116] S402. The electronic device determines a feature score based on the score of each state parameter, the weight of each state parameter, the score of each environmental parameter, and the weight of each environmental parameter.

[0117] This application does not limit the values ​​of the weights for each state parameter and each environmental parameter, and they can be configured according to actual needs. In one possible implementation, the weight of the state parameter can be greater than the weight of the environmental parameter. For state parameters, the weight of internal resistance can be configured as the highest among the state parameter weights. For environmental parameters, the weight of ambient temperature can be configured as the highest among the environmental weights.

[0118] S402 can be implemented as follows: the electronic device determines the feature score by taking the weighted average of the scores of each state parameter and each environmental parameter.

[0119] S403. The electronic device determines the target score based on the feature score, the weight of the feature score, the first score, the weight of the first score, the second score, and the weight of the second score.

[0120] Electronic devices can determine the target score by a weighted average of the feature score, the first score, and the second score.

[0121] This application does not limit the values ​​of the feature score weights, the first score weights, and the second score weights in its embodiments; these values ​​can be configured according to actual needs. For example, the weight values ​​can be configured based on the accuracy of each score in practice.

[0122] S403 can be implemented as follows: The electronic device determines the target score by using a weighted average method based on the feature score, the weight of the feature score, the first score, the weight of the first score, the second score, and the weight of the second score.

[0123] Based on the above technical means, when determining the target score, the scores of each state parameter and each environmental parameter are considered, which improves the accuracy of the target score. Furthermore, weights are assigned to each state parameter and each environmental parameter, and the target score can be adjusted by adjusting the weights, which further improves the accuracy and flexibility of the target score.

[0124] The battery state determination method provided in this application embodiment may also include, but is not limited to, the output of the target result.

[0125] refer to Figure 5 The process may include, but is not limited to, S106 and S107 as described below.

[0126] S106. If the battery status in the target result is abnormal, the electronic device determines the target abnormality type and target abnormality level of the battery.

[0127] The target anomaly type may include, but is not limited to, one or more of the following: current parameter anomaly, voltage parameter anomaly, internal resistance parameter anomaly, etc.

[0128] Anomaly level is used to characterize the degree of abnormality in the battery state. This application does not limit the method of classifying anomaly levels or the number of levels, etc., and can be configured according to actual needs.

[0129] Anomaly levels can be categorized based on target scores. For example, in a score characterizing a battery's state as abnormal, the anomaly score can be divided into multiple anomaly levels.

[0130] S106 can be implemented as follows: The electronic device first determines whether the battery status in the target result is abnormal or normal. If the battery status in the target result is abnormal, the electronic device determines the target abnormality type and target abnormality level of the battery. If the battery status in the target result is normal, the electronic device outputs a normal target result.

[0131] S107. The electronic device outputs the target anomaly type, target anomaly level, and warning information.

[0132] The warning information is used to alert users to anomalies. This application embodiment does not limit the output type or content of the warning information; it can be configured according to actual needs.

[0133] Here, the target anomaly type, target anomaly level, and output type of early warning information can be the same or different.

[0134] S107 can be implemented as follows: the electronic device outputs the target result, and outputs the target anomaly type, target anomaly level and warning information.

[0135] Based on the above technical means, it is also possible to output the target anomaly type, target anomaly level, and early warning information, thereby improving the richness of the output and the user experience.

[0136] The process of the electronic device outputting warning information in S107 will be explained below.

[0137] This process may include, but is not limited to, the following cases 1 to 3.

[0138] Scenario 1: When the target anomaly level is Level 1, output the first warning information through visual cues.

[0139] Visual cues include warning lights, displaying warning information on screens, etc.

[0140] For example, the first warning message could include: "Please check the battery status" or "Please stop the car and check".

[0141] Scenario 2: When the target anomaly level is Level 2, the first warning message is output through visual cues and the second warning message is output through auditory cues.

[0142] Auditory cues can include: outputting audio warning messages, outputting alert sounds, such as "beep beep beep," etc.

[0143] For example, the second warning message could include: "Please check the battery status immediately" or "Please stop the car immediately and check the battery status."

[0144] Scenario 3: When the target anomaly level is level three, the first warning message is output through visual cues, the second warning message is output through auditory cues, and the third warning message is output to the client via network.

[0145] Network methods can include: remote transmission via vehicle network, etc.

[0146] For example, the third warning message could include: "The battery is in a dangerous state. Please move away as soon as possible and contact the manufacturer."

[0147] The degree of abnormality increases sequentially from Level 1 to Level 2 and Level 3.

[0148] Based on the aforementioned technical means, different output methods are configured for different anomaly levels. For low anomaly levels, output is only provided visually to reduce the impact of the output on other processing. For high anomaly levels, multiple output methods are used to determine the reachability of the output, thereby improving security. In short, this multi-level output method improves the output effectiveness.

[0149] The process by which the electronic device in S102 determines the state characteristics of the battery based on state parameters and environmental parameters will be explained below.

[0150] refer to Figure 6 The process may include, but is not limited to, S601 and S602 described below.

[0151] S601, the electronic equipment determines the rate of change of each state parameter and the rate of change of each environmental parameter.

[0152] The methods for determining the rate of change may be the same or different for different parameters.

[0153] S601 can be implemented as follows: For each state parameter and each environmental parameter, the electronic device determines the rate of change of each state parameter and the rate of change of each environmental parameter respectively.

[0154] S602. The electronic device determines the state characteristics of the battery based on the rate of change of each state parameter and the rate of change of each environmental parameter.

[0155] S602 can be implemented as follows: The electronic device processes the rate of change of each state parameter through a feature extraction algorithm to obtain the feature corresponding to each state parameter, processes the rate of change of each environmental parameter to obtain the feature corresponding to each environmental parameter, and then combines these features to obtain the state features of the battery.

[0156] Based on the aforementioned technical methods, when determining the state characteristics of a battery, the characteristics are not directly determined based on parameters. Instead, the rate of change of each state parameter and each environmental parameter is determined separately. The battery's state characteristics are then determined based on these rates of change. Since the rate of change better reflects the battery's dynamic performance, the obtained state characteristics are more consistent with reality, improving the accuracy of the state characteristics, the accuracy of the target results, and the battery's safety.

[0157] In this embodiment of the application, outliers can also be removed after obtaining the rate of change of each parameter.

[0158] refer to Figure 7 The process may include, but is not limited to, S701 and S702 as shown below.

[0159] S701, the electronic device removes outliers from the rate of change of each state parameter based on a first method.

[0160] The first approach includes at least one of the following: moving average algorithm, autocorrelation function, and wavelet transform.

[0161] Moving average algorithms calculate the mean of data over a continuous time window, allowing for observation of overall data trends, reducing short-term fluctuations, highlighting long-term trends, and improving data stability. By calculating the moving average and comparing it with current data, abnormal deviations can be identified, thus eliminating outliers.

[0162] The autocorrelation function can measure the similarity of signals under different time lags, helping to identify the periodicity of data. In battery current analysis, the autocorrelation function is used to detect the existence of specific periodic changes, helping to identify abnormal data.

[0163] Wavelet transform can decompose signals into different scales, giving them high resolution in both the time and frequency domains. In battery temperature fluctuation analysis, wavelet transforms can be used to identify short-term abrupt changes, such as localized overheating, and long-term trends, such as battery aging, thus more accurately identifying abnormal data.

[0164] Wavelet transform can capture abrupt changes in signals, i.e., anomalies or fault points. Using wavelet transform to detect abrupt changes in current signals, it can effectively remove high-frequency noise while retaining useful low-frequency information, thus improving data quality.

[0165] S701 can be implemented as follows: the electronic device removes outliers from the rate of change of each state parameter based on at least one of the first methods.

[0166] S702. Electronic devices use a moving average algorithm to remove outliers from the rate of change of each environmental parameter.

[0167] S702 can be implemented as follows: the electronic device obtains the average value based on the moving average algorithm, and identifies the rate of change of environmental parameters that deviate significantly from the average value as outliers and removes them.

[0168] Based on the above technical means, different methods for removing outliers are configured according to the characteristics of different parameters, which effectively improves the accuracy of outlier removal.

[0169] The following describes the process by which the electronic device in S101 collects the battery's state parameters and the environmental parameters of the battery. When the state parameters include battery temperature, voltage, current, internal resistance, and state of charge, and the environmental parameters include ambient temperature, refer to... Figure 8 The process may include, but is not limited to, S801 and S802 described below.

[0170] S801, The electronic device collects the battery voltage and current based on a first frequency.

[0171] The embodiments of this application do not limit the value of the first frequency, and can be configured according to actual needs.

[0172] S801 can be implemented as follows: The electronic device configures the sampling frequency in the voltage and current acquisition device to a first frequency, and then acquires the voltage and current values ​​of the battery over a period of time.

[0173] S802, the electronic device collects the battery temperature, internal resistance, state of charge, and ambient temperature of the battery based on a second frequency.

[0174] The embodiments of this application do not limit the value of the second frequency, and can be configured according to actual needs.

[0175] S802 can be implemented as follows: the electronic device is configured with a second frequency in the battery temperature, internal resistance, state of charge and ambient temperature acquisition device of the battery, and then the battery temperature, internal resistance, state of charge and ambient temperature are acquired.

[0176] Based on the aforementioned technical methods, different frequencies are used for data acquisition depending on the characteristics of different parameters. This ensures the effectiveness of data acquisition while being applicable to the specific features of each parameter. For example, voltage and current parameters change rapidly, so a higher acquisition frequency is needed. Temperature parameters, on the other hand, generally do not change abruptly, so a slightly lower acquisition frequency can be used. This improves the accuracy of parameter acquisition, the accuracy of target results, and battery safety.

[0177] The process of the electronic device in S801 acquiring the battery voltage and current based on the first frequency will be explained below.

[0178] In one possible implementation, refer to Figure 9 The process may include, but is not limited to, S901 to S903 described below.

[0179] S901, Electronic equipment determines the operating status of the battery.

[0180] The operating states here may include, but are not limited to: first load state and second load state.

[0181] The load in the first load state is higher than the load in the second load state.

[0182] Therefore, the first load state can also be called the high load state, and the second load state can also be called the low load state.

[0183] S901 can be implemented as follows: the electronic device detects the current load and determines the battery's operating state based on the load.

[0184] S902. If the battery is in the first load state, the electronic device increases the first frequency; the battery voltage and current are collected based on the increased first frequency.

[0185] The embodiments of this application do not limit the magnitude of the increase in the first frequency, and can be configured according to actual needs. For example, it can be determined based on the load.

[0186] S902 can be implemented as follows: if the battery is in the first load state, the change in current and voltage is relatively large, so the electronic device increases the first frequency for collecting voltage and current; the battery voltage and current are collected based on the increased first frequency.

[0187] S903. If the battery is in the second load state, the electronic device reduces the first frequency and collects the battery voltage and current based on the reduced first frequency.

[0188] S903 can be implemented as follows: if the battery is in the second load state, the change in current and voltage is relatively small, the electronic device reduces the first frequency, and the battery voltage and current are collected based on the reduced first frequency.

[0189] The embodiments of this application do not limit the magnitude of the reduction in the first frequency, and can be configured according to actual needs. For example, it can be determined based on the load.

[0190] Based on the above technical means, the acquisition frequency can be adjusted according to the battery's working state, which further improves the data acquisition efficiency, the accuracy of the target results, and the safety of the battery.

[0191] The following example, using an electric vehicle battery, illustrates the process of determining the battery state through a specific embodiment.

[0192] Battery monitoring methods primarily rely on simple threshold detection, such as setting fixed thresholds for parameters like voltage, temperature, and current. Once these thresholds are exceeded, the system issues an alarm. However, this static monitoring approach may have the following problems: alarm response delays, fixed thresholds may lead to false alarms or missed alarms, there is a lack of in-depth analysis of the overall battery condition, and potential risks may not be identified in a timely manner.

[0193] Therefore, there is an urgent need for a more intelligent, comprehensive, and real-time battery monitoring solution to improve the accuracy and timeliness of battery anomaly detection. In recent years, with the development of big data and machine learning technologies, it has become increasingly possible to apply these technologies to battery status monitoring. By analyzing historical and real-time data, abnormal battery patterns can be identified more effectively, thereby providing early warnings.

[0194] In summary, developing an algorithm-based early warning method for electric vehicle batteries can not only improve battery safety and reliability but also provide users with a better driving experience, which is of great significance for promoting the further development of electric vehicles.

[0195] To address the aforementioned technical problems, this embodiment provides a method for early warning of abnormal electric vehicle battery conditions. The technical solution mainly includes the following steps:

[0196] Step 1, Data Acquisition: High-precision sensors are used to collect data such as battery voltage, current, temperature, state of charge (SOC), and internal resistance in real time. The collected data is transmitted to the processing device in real time via a wireless communication module.

[0197] Step 2, Data Preprocessing: The collected data is denoised and standardized. External noise is removed by filtering algorithms to ensure the accuracy and reliability of the data.

[0198] Step 3, Feature Extraction: Use time series analysis to extract the characteristic parameters of the battery state. The features of the data include temperature fluctuation rate, charge and discharge rate, internal resistance change, etc. Construct feature vectors to facilitate subsequent model training and anomaly detection.

[0199] Step 4, Algorithm Model Processing: Apply machine learning algorithms, such as random forest, support vector machine, or deep learning models, to train normal and abnormal battery behaviors and establish an anomaly detection model. This model learns from historical battery data and can automatically identify potential abnormal patterns.

[0200] Step 5, Anomaly Detection and Early Warning: The newly collected data is input into the algorithm model in real time. The model determines whether the battery status is abnormal based on the feature vector. If an anomaly is detected, the system will immediately generate an early warning message to prompt the user to take necessary measures, such as stopping the car for inspection or replacing the battery.

[0201] Step 6, Feedback Mechanism: Users can provide feedback on the system's early warning results. The system will continuously optimize the algorithm model based on user feedback to improve the accuracy and reliability of anomaly detection.

[0202] This embodiment provides an electric vehicle battery anomaly warning device, referencing... Figure 10 The electric vehicle battery abnormality warning device 100, as shown, may include:

[0203] The data acquisition module 1001 collects real-time battery-related data.

[0204] The data preprocessing module 1002 performs noise reduction and standardization on the collected data.

[0205] The feature extraction module 1003 extracts the feature parameters of the battery state and constructs a feature vector.

[0206] The anomaly detection module 1004 uses machine learning algorithms to determine the battery status and generate warning information.

[0207] User feedback module 1005 optimizes the algorithm model based on user feedback on early warnings.

[0208] This embodiment provides a method for early warning of abnormalities in electric vehicle batteries, including: data collection through various high-precision sensors installed on the battery pack, including temperature sensors, voltage sensors, current sensors, and internal resistance measuring devices; different collection frequencies are set for different parameters according to requirements, rapidly changing data such as voltage and current can be set to high-frequency collection, while relatively stable data such as temperature can be set to low-frequency collection; the system can dynamically adjust the collection frequency according to the battery's operating state, increasing the collection frequency under high load or rapid charging / discharging conditions, and decreasing the collection frequency under normal and stable operating conditions.

[0209] Data is transmitted to the central processing module via a bus, and simultaneously transmitted to the cloud via a wireless communication module for subsequent analysis and model training. After data acquisition, the raw data undergoes preprocessing, primarily including the following steps: noise reduction (Kalman filtering is used for time-series data such as voltage and current); moving average filtering is used to eliminate occasional abrupt changes in relatively stable data such as temperature and state of charge, and adjacent data are weighted and averaged; data integrity checks and missing value imputation are performed using interpolation or regression models based on historical data, and outliers are removed or replaced by setting thresholds; data format standardization and unit standardization are implemented; current and voltage data are normalized to a range of 0-1; feature dimensionality reduction is performed on high-dimensional datasets using principal component analysis or linear discriminant analysis algorithms; the preprocessed dataset is then divided into training, validation, and test sets.

[0210] The feature extraction process will be explained below.

[0211] Input data is processed, and feature extraction is performed on the collected data. Feature extraction mainly includes selecting key features, constructing feature vectors, weighted processing of features, and applying time series analysis. After determining the key features, the system further optimizes the features through feature engineering to improve the model's discriminative ability. Feature engineering includes constructing composite features, weighted processing, and feature derivation.

[0212] refer to Figure 11 The process may include, but is not limited to, S1101 to S1106 described below.

[0213] S1101, Input preprocessed data.

[0214] S1102, Calculate temperature fluctuation rate using the sliding window method.

[0215] Temperature fluctuation rate describes the magnitude of temperature change per unit time, assuming a time interval of [missing information]. The temperature fluctuation rate can be determined by the following formula (2).

[0216] Formula (2);

[0217] In formula (2), N represents the number of sampling points within the time interval. Indicates the interval time. This represents the temperature at time i; Indicates in Temperature at any moment; This represents the temperature fluctuation rate.

[0218] Temperature fluctuation rate can help the system identify abnormal temperature fluctuations, thereby providing early warnings of risks such as battery overheating.

[0219] Voltage and current characteristics: By calculating the rate of change of voltage or current per unit time, the charge and discharge rate of the battery can be obtained. Abnormal changes in the charge and discharge rate may be a signal that the battery is in an undesirable state such as overcharging or over-discharging.

[0220] For example, the rate of change of voltage current can be determined by the following formula (3).

[0221] , Formula (3);

[0222] In formula (3), and These represent the changes in voltage and current per unit time, i.e., the rates of change of voltage and current, respectively. Indicates a unit of time; and These represent the rate of change of voltage and the rate of change of current, respectively.

[0223] Collect the maximum and minimum values ​​of voltage and current within a time window to analyze their fluctuations. Significant differences between the peaks and valleys of voltage or current may indicate instability in the battery.

[0224] S1103, Calculate the rate of change of internal resistance.

[0225] The rate of change of internal resistance reflects the health status of the battery and is an important characteristic in the battery degradation process. For example, the rate of change of internal resistance can be determined by formula (4).

[0226] Formula (4);

[0227] In formula (4), Indicates the rate of change of internal resistance. This represents the internal resistance at the current moment. This indicates the internal resistance at the previous moment. It represents a time interval at a given moment.

[0228] The internal resistance is smoothed using a time-weighted average, and the trend of internal resistance change is calculated to identify chronic battery degradation. If the rate of change of internal resistance continues to increase, the system can determine that the battery is in a degraded state.

[0229] S1104. Calculate the rate of change of SOC.

[0230] Average consumption rate is the rate of change of SOC within a time window to estimate the battery's power consumption under the current load.

[0231] Instantaneous rate of change: At each sampling point, the instantaneous rate of change of SOC is calculated to monitor the real-time discharge state of the battery;

[0232] Outliers can also be removed in the following ways.

[0233] Moving average is used to smooth characteristics such as voltage, current, and temperature, reducing short-term noise fluctuations. The formula for calculating the moving average can be found in the following formula (5).

[0234] Formula (5);

[0235] In formula (5), Let k represent the moving average at time t, and k be the window size. Let be the eigenvalue at time t.

[0236] In this case, the appropriate window size can usually be selected based on the data fluctuation.

[0237] The autocorrelation function is used to evaluate the correlation of a feature under different time delays. It can identify the periodic or trend characteristics of battery parameters. For the current feature I, its autocorrelation coefficient can be determined by formula (6).

[0238] = Formula (6);

[0239] In formula (6), Represents the autocorrelation coefficient. Here, k represents the average current value, N represents the number of lag steps, and N represents the total number of steps. This represents the current value at time t.

[0240] Wavelet transform is used to decompose temperature signals for battery characteristics with multi-scale properties, such as battery temperature fluctuations, in order to detect abnormal characteristics of the battery at different time scales.

[0241] Wavelet transform can be implemented using the following formula (7).

[0242] Formula (7);

[0243] In formula (7), x(t) represents the original signal; Let a and b represent the wavelet basis functions, respectively, and a and b be the scaling and translation parameters. This represents the signal after transformation.

[0244] The raw signal here can be the raw signal of temperature, the raw signal of voltage, etc.

[0245] S1105, Feature vector construction.

[0246] The feature vector can be constructed using the following formula (8).

[0247] Formula (8);

[0248] In formula (8), Represents the constructed features. Indicates the rate of temperature change. Indicates the rate of change of voltage. Indicates the rate of change of current. Indicates the internal resistance transformation rate. This represents the rate of change of SOC.

[0249] For different features in the feature vector, the system weights them according to their importance to anomaly detection. For example, the rate of change of battery internal resistance has a greater impact on battery degradation, so it is given a higher weight.

[0250] S1106, Output feature vector.

[0251] The feature extraction system can extract key features reflecting the health status of the battery from the raw data, providing high-quality data input for anomaly detection and enabling accurate battery status monitoring and risk warning.

[0252] The model processing procedure is explained below.

[0253] This embodiment uses a combination of deep learning-based Long Short-Term Memory (LSTM) networks and Random Forest models to process time series data and perform multi-feature comprehensive analysis, respectively. Since battery data has time-series characteristics and multiple parameters such as voltage, current, and temperature are interrelated, the above two models are selected.

[0254] refer to Figure 12 The processing of the model, as shown, may include, but is not limited to, S1201 to S1206 below.

[0255] S1201. Data Preparation: Data Collection, Labeling, and Division.

[0256] Collect various state parameters of the battery and label them manually or automatically to mark abnormal and normal battery states. The labeling method is based on battery laboratory test data, simulation data and actual operation data. Divide the dataset into training set, validation set and test set. Usually, an 8:1:1 ratio is used to ensure that the test set is sufficiently independent and can effectively verify the generalization ability of the model.

[0257] S1202, Model Selection and Construction.

[0258] The LSTM model, used for time series data processing, has a basic structure consisting of an input layer, multiple LSTM hidden layers, and an output layer. The specific structural design is as follows:

[0259] The input layer receives time-series data from the battery, such as temperature, current, and SOC. Assuming the time window is T, the input data format is (X(tT), ..., Xt), where X is a feature vector. Two LSTM hidden layers are set, with the number of layers depending on the amount of data and the complexity requirements of the model. Each hidden layer has 128 neurons and uses the ReLU activation function to improve the non-linear expressive power of the model. The output layer is a single neuron with the Sigmoid activation function, and the output predicted value is a probability value between 0 and 1, used to determine whether there is an anomaly within the time window. The key parameters of the LSTM model include the time window size (e.g., 10 seconds), the number of hidden layers (e.g., 2 layers), and the number of neurons (e.g., 128). The optimal configuration is found through parameter tuning.

[0260] The Random Forest model is used to process multi-dimensional static features and determine whether a battery is abnormal under a specific state. The model's structure is designed as follows:

[0261] The input is a preprocessed feature vector, and each feature is standardized. The random forest model contains 100 decision trees (which can be adjusted according to the amount of data and performance requirements). Each tree is trained using a random subset of features to increase the model's generalization ability. The random forest model outputs the final judgment result through a majority voting mechanism. When the majority of trees judge it as abnormal, an abnormal alarm is output; otherwise, it is judged as normal.

[0262] S1203, Model Training: Training based on training set data.

[0263] To minimize prediction error, the parameters are tuned through model training. The training process includes loss function selection, parameter optimization, and model validation.

[0264] The loss function of the LSTM model is to use the binary cross-entropy loss function to calculate the prediction error. The binary cross-entropy loss function can be referred to in the following formula (9).

[0265] Suppose the loss function is defined as:

[0266] Formula (9);

[0267] In formula (9), Let N represent the loss value, N represent the number of training data points, and y represent the true label. This represents the predicted value output by the model, where i represents the index of the current data.

[0268] By minimizing the loss function, the LSTM model can more accurately identify abnormal states.

[0269] Random forest models are optimized based on classification accuracy to improve the model's classification accuracy on the validation set. Since random forest is a non-parametric model, the optimization process mainly improves performance by increasing the number of trees and the randomness of features.

[0270] S1204, Hyperparameter Optimization.

[0271] Hyperparameter optimization of LSTM models: Key parameters of LSTM models include time window size, number of hidden layers, and number of neurons. A combination of grid search and cross-validation is used to train the model under different parameter combinations, and the configuration that performs best on the validation set is selected.

[0272] Random forest model parameter optimization: The main parameters of random forest include the number of decision trees, maximum depth, and number of feature selections. A random search algorithm is used to find the optimal configuration in the parameter space to improve the classification accuracy of the model.

[0273] S1205, Model Validation.

[0274] After the model is trained, it is validated using a test set to evaluate its performance. The evaluation metrics used include accuracy, precision, recall, and score. In particular, to ensure the effectiveness of the model in anomaly detection, recall is the focus, in order to reduce the probability of false negatives.

[0275] S1206, Model Deployment and Application.

[0276] After training the LSTM and random forest models, they are integrated together to build the final anomaly detection system.

[0277] The integrated logic uses the LSTM model to detect short-term anomalies in time series data. When the LSTM model outputs an anomaly alert, it directly generates a warning. The random forest model is used to further analyze multi-dimensional static features and output a comprehensive judgment result, ensuring that the random forest can supplement the detection when the LSTM model misses a detection.

[0278] Real-time optimization: To improve the real-time performance of the system, the inference process of LSTM and random forest models is parallelized; by deploying the models on edge computing devices, they can efficiently process real-time data.

[0279] This embodiment effectively improves the accuracy and response speed of anomaly detection by combining multi-model detection, early warning level classification, and intelligent prompting mechanisms.

[0280] The anomaly detection process will be explained below.

[0281] The anomaly detection process includes several key steps: data input, model analysis, anomaly identification, and result output. (Reference) Figure 13The process may include, but is not limited to, S1301 to S1309 below.

[0282] S1301, Anomaly detection begins.

[0283] S1302, Feature extraction module inputs feature vector.

[0284] The system acquires various battery parameters in real time from the data acquisition module, including voltage, current, temperature, internal resistance, and SOC. After preprocessing, these data are input into the LSTM and random forest models for time series detection and multidimensional feature detection, respectively.

[0285] S1303 and LSTM models are used to analyze and process time series data.

[0286] LSTM model detection involves analyzing the trends of time series data such as temperature and voltage. The model predicts the values ​​of various parameters at the next moment and compares them with the actual collected values. If the error exceeds a preset threshold, it is judged as abnormal.

[0287] S1304, Random Forest Model Analysis and Processing of Multidimensional Feature Data.

[0288] Random forest model detection involves analyzing the multidimensional features of the input to determine whether the current state is abnormal. If the majority of decision trees determine that it is abnormal, then an abnormal result is output.

[0289] S1305, Calculate the comprehensive anomaly score, combining the model output with the feature score.

[0290] Anomaly fusion judgment: When the output of either the LSTM model or the random forest model is abnormal, the system will further fuse the results of the two models and give a comprehensive score.

[0291] S1306, Determine the level of abnormality.

[0292] Whether an alert is triggered is determined based on a comprehensive score, which is calculated by weighting the confidence level of the model's output anomaly.

[0293] To improve the usability of the early warning information, this embodiment classifies the abnormal detection results into different levels to correspond to different battery states and corresponding countermeasures. The early warning levels include the following three:

[0294] Level 1 Warning (Minor Anomaly): A minor anomaly is detected that is not enough to pose a direct threat to battery safety. It may be caused by environmental changes or occasional noise. The system will record this status and display a prompt message to the user, such as "Battery status is normal, but please remain vigilant."

[0295] Level 2 Warning (Moderate Anomaly): A significant anomaly has been detected, potentially posing a safety hazard. At this time, the system will issue a suggestion to the user, such as "Battery status is abnormal; please check or contact the service center as soon as possible."

[0296] Level 3 Warning (Serious Anomaly): A serious anomaly is detected, indicating a potential battery malfunction or safety risk. The system will immediately issue an alarm sound, prompting the user that "the battery has a serious malfunction. Please stop the vehicle immediately for inspection and ensure safety."

[0297] To improve the accuracy of early warnings, this embodiment employs a scoring mechanism, which scores anomaly detections based on the degree of anomaly of each feature and the model's confidence level. The scoring mechanism includes:

[0298] Single feature scoring: For each feature (such as temperature, voltage, etc.), a score is given based on its deviation from the normal range. The single feature score can be obtained according to the above formula (1).

[0299] The model confidence score is calculated by combining the anomaly confidence scores of the LSTM and Random Forest models with the scores of the single features. The comprehensive anomaly score can be obtained according to the following formula (10).

[0300] Formula (10).

[0301] In formula (10), , and These are weighting coefficients, used to reflect the impact of different models and features on the final judgment. The score output by LSTM. The score for the output of the random forest. The score for the feature corresponding to each parameter.

[0302] S1307. Generate warning information with warning type and operation suggestions.

[0303] When the overall score reaches the warning trigger condition, the system will generate corresponding prompt information based on the warning level. The generated warning information includes the following:

[0304] Warning type: Displays the type of abnormality, such as abnormal temperature or abnormal current.

[0305] The level of abnormality is indicated by a Level 1, Level 2, or Level 3 warning, based on the scoring results.

[0306] Operation suggestions are provided, such as "Please check the battery status" or "Please stop the car immediately and check."

[0307] The alert time displays the time when the alert occurred, making it easier for users to track historical anomalies.

[0308] S1308, visual, auditory, and remote prompts for users.

[0309] To ensure users are promptly notified of any abnormal situations, the system employs a multi-level early warning mechanism, including visual, auditory, and remote notification methods:

[0310] Visual cues display warning information on the vehicle's in-vehicle display screen, including the level of abnormality, type of abnormality, and operational suggestions. These visual cues are non-mandatory reminders and can be viewed selectively by the user.

[0311] Auditory alerts: When the warning level reaches level two or above, the system will issue an audio alarm to remind the user to check the battery status in time.

[0312] Remote notifications: The system will send abnormal information to the cloud via the vehicle network and push it to the user's mobile application, so that the user can know the battery status in time even when outside the vehicle. In the case of a level 3 warning, the system will prioritize sending an emergency push notification.

[0313] S1309. Record abnormal events and upload them to the cloud.

[0314] Warning history records: All warning records will be uploaded to the cloud for subsequent model optimization and battery health management analysis.

[0315] After an alert is generated, the system records the time, type, and processing result of the anomaly detection and allows users to provide feedback. The user feedback module helps the system collect the following information: anomaly confirmation, where users can manually confirm whether it is a genuine anomaly to optimize the model's false alarm rate; and anomaly cause, where users can select the environment in which the anomaly occurred (e.g., high temperature environment, long-distance driving, etc.) to help the system adaptively adjust in similar situations.

[0316] Through the anomaly detection and early warning mechanism provided in this embodiment, the system can accurately identify battery anomalies of different degrees and provide graded prompts to users according to the anomaly situation, ensuring vehicle safety and user experience.

[0317] This embodiment has the following technical effects:

[0318] 1. By combining Long Short-Term Memory (LSTM) network and Random Forest model with machine learning algorithms, dynamic learning and adaptation are achieved. The system can self-adjust when the battery state changes, which can improve the accuracy and robustness of anomaly detection and make up for the shortcomings of a single model in specific scenarios.

[0319] 2. Employing multiple feature extraction techniques, such as volatility calculation and internal resistance change rate, to comprehensively analyze battery status; this comprehensive feature extraction method improves the effectiveness of anomaly detection and ensures the system's adaptability under different operating conditions.

[0320] 3. A dynamic scoring mechanism is introduced, which quantifies the battery status by comprehensively considering the deviation of various features and the confidence of the model. This method not only improves the flexibility of anomaly identification, but also effectively distinguishes different types of anomalies, thereby optimizing the early warning strategy.

[0321] 4. By utilizing a cloud computing platform to analyze and process collected historical data, intelligent battery health management and anomaly detection can be achieved. Through cloud services, users can remotely monitor battery status, obtain real-time early warning information, and improve the safety of electric vehicles.

[0322] 5. An early warning level classification mechanism and multi-level early warning prompts are introduced, dividing abnormal detection results into three levels of early warning, and providing corresponding operation suggestions and prompts according to different levels of abnormality; through a multi-level early warning prompt mechanism including visual, auditory and remote notifications, users can be informed of abnormal battery status in a timely manner in various situations, which improves users' trust in the system and can quickly guide users to take action in high-risk situations.

[0323] This embodiment of the battery anomaly early warning system can effectively identify various potential risks such as abnormal temperature and increased internal resistance, and issue early warning information in advance to avoid safety accidents caused by battery failure. The system can be widely used in various types of electric vehicles, including passenger cars, commercial vehicles and electric buses.

[0324] The electric vehicle battery anomaly early warning method provided in this embodiment, by combining advanced machine learning algorithms, can intelligently monitor battery status, identify potential anomalies in real time, and improve battery safety and reliability. With the continuous expansion of the electric vehicle market, the implementation of this embodiment will effectively enhance the overall safety level of electric vehicles, providing users with a safer and more reliable driving experience.

[0325] Secondly, embodiments of this application provide a battery state determination device. This battery state determination device can be deployed in an electronic device. (See reference...) Figure 14 The battery status determination device 140 shown may include, but is not limited to: acquisition unit 1401, first determination unit 1402, first analysis unit 1403, second analysis unit 1404, and second determination unit 1405.

[0326] The acquisition unit 1401 is used to acquire the battery's state parameters and the environmental parameters in which the battery is located;

[0327] The first determining unit 1402 is used to determine the state characteristics of the battery based on state parameters and environmental parameters;

[0328] The first analysis unit 1403 is used to perform anomaly analysis on the state characteristics in the time dimension using the first model to obtain the first result of the battery; the first result is used to characterize whether the battery is abnormal in the time dimension.

[0329] The second analysis unit 1404 is used to perform spatial dimension anomaly analysis on the state characteristics through the second model to obtain the second result of the battery; the second result is used to characterize whether the battery is abnormal in the spatial dimension.

[0330] The second determining unit 1405 is used to determine a target result for characterizing the state of the battery based on the first result and the second result.

[0331] In some embodiments, the second determining unit 1405 is further configured to: if the first result and the second result are the same, determine the state of the battery as the first result; if the first result and the second result are different, determine a target score based at least on the first score and the second score; determine a target result based on the target score; wherein the first score is the score output by the first model; and the second score is the score output by the second model.

[0332] In some embodiments, the second determining unit 1405 is further configured to: determine a target score based on a first score and a second score; and determine a target score based on a first score, a second score, and a feature score.

[0333] In some embodiments, a score for each state parameter and a score for each environmental parameter are determined; a feature score is determined based on the score for each state parameter, the weight of each state parameter, the score for each environmental parameter, and the weight of each environmental parameter; and a target score is determined based on the feature score, the weight of the feature score, a first score, the weight of the first score, a second score, and the weight of the second score.

[0334] In some embodiments, the battery state determination device 140 may further include an output unit, the output unit being used for:

[0335] If the battery status is abnormal in the target result, determine the target abnormality type and target abnormality level of the battery; output the target abnormality type, target abnormality level and warning information.

[0336] In some embodiments, the output unit is further configured to: output a first warning message via visual cues when the target anomaly level is Level 1; output the first warning message via visual cues and the second warning message via auditory cues when the target anomaly level is Level 2; output the first warning message via visual cues and the second warning message via auditory cues when the target anomaly level is Level 3, and output the third warning message to the client via a network; wherein the anomaly severity of Level 1, Level 2, and Level 3 anomalies increases sequentially.

[0337] In some embodiments, the first determining unit 1402 is further configured to: determine the rate of change of each state parameter and the rate of change of each environmental parameter respectively; and determine the state characteristics of the battery based on the rate of change of each state parameter and the rate of change of each environmental parameter.

[0338] In some embodiments, the first determining unit 1402 is further configured to: remove outliers from the rate of change of each state parameter based on a first method; the first method includes at least one of a moving average algorithm, an autocorrelation function, and a wavelet transform; and remove outliers from the rate of change of each environmental parameter based on a moving average algorithm.

[0339] In some embodiments, the acquisition unit 1401 is further configured to perform the following when the state parameters include battery temperature, voltage, current, internal resistance, and state of charge, and the environmental parameters include ambient temperature: acquiring the battery voltage and current based on a first frequency; and acquiring the battery temperature, internal resistance, state of charge, and ambient temperature based on a second frequency.

[0340] In some embodiments, the operating state of the battery is determined; if the operating state of the battery is a first load state, the first frequency is increased; the voltage and current of the battery are collected based on the increased first frequency; if the operating state of the battery is a second load state, the first frequency is decreased, and the voltage and current of the battery are collected based on the decreased first frequency; wherein the load of the first load state is higher than the load of the second load state.

[0341] Thirdly, this application also provides an electronic device, which includes a processor and a memory, wherein the memory stores a computer program or instructions, which, when executed by the processor, implement the method provided in the first aspect.

[0342] Fourthly, embodiments of this application provide a storage medium, namely a computer-readable storage medium, on which a computer program or instructions are stored, which, when executed by a processor, implement the method provided in the first aspect above.

[0343] Fifthly, embodiments of this application provide a computer program product, which includes a computer program or instructions. When the computer program or instructions are executed by a processor, they implement the method provided in the first aspect described above.

[0344] It should be noted that the descriptions of the above embodiments of storage media, devices, and program products are similar to the descriptions of the above method embodiments, and have similar beneficial effects. For technical details not disclosed in the embodiments of storage media, devices, apparatus, and program products of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0345] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment" or "in some embodiments" appearing throughout the specification do not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of this application, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the above-described embodiments are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0346] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0347] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0348] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0349] In addition, each functional unit in the various embodiments of this application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0350] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory (ROM), magnetic disks, or optical disks.

[0351] Alternatively, if the integrated units described above are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to related technologies, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0352] The above description is merely an embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method of determining a state of a battery, characterized by, The method comprises: collecting state parameters of a battery and environmental parameters in which the battery is located; determining state features of the battery based on the state parameters and the environmental parameters; inputting the state features of the battery into a first model to perform time-dimension anomaly analysis on the state features by the first model to obtain a first result of the battery; the first result is used to represent whether the battery is abnormal in the time dimension; inputting the state features of the battery into a second model to perform space-dimension anomaly analysis on the state features by the second model to obtain a second result of the battery; the second result is used to represent whether the battery is abnormal in the space dimension; if the first result and the second result are the same, determining that the state of the battery is the first result; if the first result and the second result are different, determining a target score based on at least a first score and a second score; and determining a target result based on the target score; wherein the first score is a score converted from an abnormal probability output by the first model; the second score is a score converted from an abnormal voting ratio output by the second model; and the target score is determined based on at least the first score and the second score, at least including determining an average value or a weighted average value of the first score and the second score as the target score.

2. The method of claim 1, wherein, The target score is determined based on at least the first score and the second score, including: determining a score of each state parameter and a score of each environmental parameter; determining a feature score based on the score of each state parameter, a weight of each state parameter, the score of each environmental parameter, and a weight of each environmental parameter; determining the target score based on the feature score, a weight of the feature score, the first score, a weight of the first score, the second score, and a weight of the second score.

3. The method according to claim 1 or 2, characterized in that, The method further comprises: if the state of the battery in the target result is abnormal, determining a target abnormal type and a target abnormal level of the battery; outputting the target abnormal type, the target abnormal level, and warning information.

4. The method of claim 3, wherein, Outputting the warning information includes: in a case where the target abnormal level is a first abnormal level, outputting first warning information through a visual prompt mode; in a case where the target abnormal level is a second abnormal level, outputting the first warning information through the visual prompt mode and outputting second warning information through an auditory prompt mode; in a case where the target abnormal level is a third abnormal level, outputting the first warning information through the visual prompt mode, outputting the second warning information through the auditory prompt mode, and outputting third warning information through a network mode to a client; wherein the abnormal degrees of the first abnormal level, the second abnormal level, and the third abnormal level increase in turn.

5. The method according to claim 1 or 2, characterized in that, The determination of the state features of the battery based on the state parameters and the environmental parameters includes: determining a change rate of each state parameter and a change rate of each environmental parameter; determining the state features of the battery based on the change rate of each state parameter and the change rate of each environmental parameter.

6. The method of claim 5, wherein, Before performing the determination of the state feature of the battery based on the rate of change of each of the state parameters and the rate of change of each of the environment parameters, the method further comprises: removing outliers in the rate of change of each of the state parameters based on a first manner; the first manner comprises at least one of moving average algorithm, autocorrelation function and wavelet transform; removing outliers in the rate of change of each of the environment parameters based on the moving average algorithm.

7. The method according to claim 1 or 2, characterized in that, In the case that the state parameters comprise battery temperature, voltage, current, internal resistance and state of charge, and the environment parameters comprise environment temperature, the collecting of the state parameters of the battery and the environment parameters in which the battery is located comprises: collecting the voltage and current of the battery based on a first frequency; collecting the battery temperature, internal resistance, state of charge of the battery and the environment temperature based on a second frequency.

8. The method of claim 7, wherein, The collecting of the voltage and current of the battery based on the first frequency comprises: determining the working state of the battery; if the working state of the battery is a first load state, increasing the first frequency; collecting the voltage and current of the battery based on the increased first frequency; if the working state of the battery is a second load state, decreasing the first frequency; collecting the voltage and current of the battery based on the decreased first frequency; wherein the load load of the first load state is higher than the load load of the second load state.

9. A battery state determination device characterized by comprising: The apparatus comprises: a collecting unit configured to collect state parameters of a battery and environment parameters in which the battery is located; a first determining unit configured to determine a state feature of the battery based on the state parameters and the environment parameters; a first analyzing unit configured to input the state feature of the battery into a first model, and perform time-dimension abnormality analysis on the state feature through the first model to obtain a first result of the battery; the first result is used to represent whether the battery is abnormal in the time dimension; a second analyzing unit configured to input the state feature of the battery into a second model, and perform space-dimension abnormality analysis on the state feature through the second model to obtain a second result of the battery; the second result is used to represent whether the battery is abnormal in the space dimension; a second determining unit configured to, if the first result and the second result are the same, determine that the state of the battery is the first result; if the first result and the second result are different, determine a target score based on at least a first score and a second score; determine a target result based on the target score; wherein the first score is a score converted from an abnormal probability output by the first model; the second score is a score converted from an abnormal voting ratio output by the second model; the determination of the target score based on at least the first score and the second score comprises at least determining an average value or a weighted average value of the first score and the second score as the target score.

10. An electronic device, comprising: The electronic device comprises a processor and a memory, and the memory has stored thereon a computer program or instructions, which, when executed by the processor, implement the method of any one of claims 1-8. The electronic device comprises a processor and a memory, and the memory has stored thereon a computer program or instructions, which, when executed by the processor, implement the method of any one of claims 1-8.

11. A computer readable storage medium, characterized in that, The storage medium stores a computer program or instructions, and the computer program or instructions are executed by a processor to implement the method in any one of claims 1-8.

12. A computer program product, characterised in that, The computer program product comprises a computer program or instructions, and the computer program or instructions are executed by a processor to implement the method in any one of claims 1-8.

Images