Method and device for predicting remaining service life of lithium battery

Abstract

The invention relates to the technical field of battery life prediction, in particular to a method and device for predicting the remaining service life of a lithium battery. According to the scheme, after multiple key parameters of the lithium battery are obtained, the weight is dynamically adjusted through the weight adjustment algorithm according to the historical change of the parameters and the correlation between the parameters and the service life, prediction is more accurate, and one-sidedness is avoided. A matrix fusing time-space and battery state characteristics is constructed, multi-dimensional information is integrated, a comprehensive data basis is provided for prediction, and accuracy and reliability are improved. And according to geographical and environmental parameters, an environment preset model or a general model is matched, so that the prediction of a specific region is met, general conditions can be coped, and the applicability of the method is enhanced. The environment preset model is pre-trained based on historical data and can be updated online, the influence of the environment on the service life of the battery is tracked in real time, timely and accurate prediction is provided, and lithium battery maintenance and replacement decisions are assisted.

Description

Technical Field

[0001] The present invention relates to the technical field of battery life prediction, and particularly relates to a method and device for predicting the remaining useful life of a lithium battery. Background Art

[0002] Due to its advantages such as high energy density, long cycle life, and low self-discharge rate, lithium batteries have been widely used in many fields such as electronic devices, electric vehicles, and energy storage systems. However, as the lithium battery is used, its performance will gradually decline and the remaining useful life will continuously decrease. Accurately predicting the remaining useful life of a lithium battery is of great significance for ensuring the safe and stable operation of equipment, optimizing the equipment maintenance plan, and reducing the use cost.

[0003] In the existing technologies for predicting the remaining useful life of lithium batteries, there are many deficiencies. On the one hand, some current methods only consider the state parameters of the battery itself, such as the remaining capacity of the lithium-ion battery, the current and voltage during the discharge process, and the number of charge-discharge cycles, while ignoring the influence of geographical location parameters and environmental parameters on the life of the lithium battery. In fact, there are differences in climatic conditions, altitude, etc. at different geographical locations, and these environmental factors will significantly affect the performance and life of the lithium battery. For example, in a high-temperature environment, the chemical reaction rate inside the lithium battery accelerates, which may lead to a rapid decline in the battery capacity; in a low-temperature environment, the internal resistance of the battery increases and the charge-discharge efficiency decreases. Therefore, if these factors are not considered, the accuracy of the prediction result will be greatly reduced.

[0004] On the other hand, the existing prediction methods often do not dynamically adjust the influence degree of different parameters on the remaining useful life of the battery. Each parameter has different influences on the battery life in different usage scenarios and time stages. If a fixed weight is used to measure each parameter, the actual state of the battery cannot be accurately reflected, thus affecting the prediction accuracy.

[0005] In summary, the existing technologies for predicting the remaining useful life of lithium batteries have many defects and are difficult to meet the requirements of high-precision prediction in practical applications. Summary of the Invention

[0006] To solve the above problems, the present invention discloses a method and device for predicting the remaining useful life of a lithium battery.

[0007] To achieve the above object, on the one hand, the present application discloses a method for predicting the remaining useful life of a lithium battery, including the following steps: Obtain the key parameters of the lithium battery during operation, where the key parameters include geographical location parameters, environmental parameters, the remaining capacity of the lithium-ion battery, the internal temperature of the battery, the current and voltage during the discharge process, the number of charge-discharge cycles, and the battery health state parameters; Monitor the key parameters and dynamically adjust the weights of each parameter based on a preset weight adjustment algorithm, where the weight adjustment algorithm determines the weight values of each parameter according to the historical change trend of the parameter and its correlation with the remaining service life of the battery; Construct an input feature matrix for the remaining service life of the lithium-ion battery, which integrates spatio-temporal features and battery state features; When the geographical location parameter and / or environmental parameter meet the preset rules, call the environmental preset model matching the target climate region to calculate and output the remaining service life of the lithium battery; otherwise, use the general model to calculate and output the remaining service life of the lithium battery; wherein, the environmental preset model is a machine learning model pre-trained with historical degradation data of the target climate region and can be updated online to adapt to environmental changes.

[0008] In the above solution, the environmental parameters include environmental temperature, humidity and atmospheric pressure data; the geographical location parameter obtains the longitude and latitude coordinates and altitude value through the GPS module.

[0009] The input feature matrix contains a three-dimensional spatio-temporal tensor associated with a timestamp, where the first dimension is the quantization value of the geographical location coordinates, the second dimension is the parameter change rate within the time window, and the third dimension contains the battery state feature vector, and the feature vector is processed by dimensionality reduction through the principal component analysis method.

[0010] The determination of the preset rules adopts a climate pattern recognition model under the transfer learning framework, specifically including: inputting the real-time geographical location parameter into the pre-trained ResNet-Transformer hybrid network to output the climate region probability distribution; when the target climate region probability value exceeds the threshold and the environmental parameter matches the typical degradation pattern, activate the corresponding environmental preset model.

[0011] On the other hand, the present application also discloses a device for predicting the remaining service life of a lithium battery, and the device includes: A parameter acquisition module for acquiring the key parameters of the lithium battery during operation, where the key parameters include geographical location parameters, environmental parameters, remaining capacity of the lithium-ion battery, internal battery temperature, current and voltage during discharge, number of charge and discharge cycles, and battery health state parameters; A parameter monitoring and weight adjustment module for monitoring the key parameters and dynamically adjusting the weights of each parameter based on a preset weight adjustment algorithm, where the weight adjustment algorithm determines the weight values of each parameter according to the historical change trend of the parameter and its correlation with the remaining service life of the battery; A feature matrix construction module for constructing an input feature matrix for the remaining service life of the lithium-ion battery, which integrates spatio-temporal features and battery state features; The remaining service life calculation module, when the geographical location parameter and / or the environmental parameter meet the preset rules, calls the environmental preset model matching the target climate region to calculate and output the remaining service life of the lithium battery; otherwise, uses the general model to calculate and output the remaining service life of the lithium battery; wherein, the environmental preset model is a machine learning model pre-trained based on the historical degradation data of the target climate region and can be updated online to adapt to environmental changes.

[0012] The output module is used to display or transmit the calculated remaining service life of the lithium battery.

[0013] In a possible implementation manner, the device further includes: The feature extraction module is used to perform learnable PCA dimensionality reduction on the battery operating parameters to generate a spatio-temporal fusion feature matrix. The incremental update module performs differential parameter update when there is an environmental mutation or the prediction error exceeds the threshold.

[0014] The feature extraction module further includes: The hash coding unit, which is used to discretize the temperature and the number of cycles into hash values of a preset length. The hash values of the preset length are used as the input of the remaining service life calculation module to calculate the remaining service life of the lithium battery.

[0015] The remaining service life calculation module includes a climate judgment module, which is used to match the corresponding climate classification code according to the geographical location coordinates. The climate judgment module performs the following operations: Maps the longitude and latitude coordinates to a climate classification code. Matches the preset model identifier through a look-up table method.

[0016] The incremental update module triggers the update process when the following conditions are met simultaneously: The environmental temperature change rate ≥ 5°C / hour; The prediction error exceeds 10% of the preset threshold for 3 consecutive times; The update data transmission adopts an incremental compression method, and the data integrity is verified through a hash check code.

[0017] In the technical solution of this application, after obtaining various key parameters of the lithium battery operation, through the preset weight adjustment algorithm, combined with the historical change trend of the parameters and the correlation with the remaining service life of the battery, the weights of each parameter are dynamically adjusted, making the prediction of the remaining service life of the lithium battery more accurate and avoiding the one-sidedness of the prediction in the single fixed weight mode.

[0018] Construct an input feature matrix that integrates spatio-temporal features and battery state features, and integrate multi-dimensional information such as time, space, and its own state during the operation of the lithium battery. This multi-feature fusion method provides a more comprehensive and representative data basis for predicting the remaining useful life of the lithium battery, thereby improving the accuracy and reliability of the prediction.

[0019] When the geographical location parameter and / or environmental parameter conform to the preset rule, call the environmental preset model matching the target climate region to calculate and output the remaining useful life of the lithium battery; otherwise, use the general model for calculation. This flexible model selection mechanism can not only make accurate predictions for specific climate regions, but also provide effective prediction results under general circumstances, greatly improving the applicability of this method under different environmental conditions.

[0020] The environmental preset model is a machine learning model pre-trained with historical degradation data of the target climate region and can be updated online to adapt to environmental changes. This enables the model to track in real time the impact of environmental changes on the life of the lithium battery, ensuring that accurate predictions of the remaining useful life of the lithium battery can still be provided in a changing environment, enhancing the timeliness and reliability of the prediction, and providing a more timely and accurate decision-making basis for the maintenance and replacement of the lithium battery. Brief Description of the Drawings

[0021] Figure 1 It is a schematic flowchart of the method for predicting the remaining useful life of the lithium battery in the embodiment of the present application; Figure 2 It is a schematic structural diagram of the device for predicting the remaining useful life of the lithium battery in the embodiment of the present application; Figure 3 It is a schematic structural diagram of the device for predicting the remaining useful life of the lithium battery in the embodiment of the present application, which includes a feature extraction module and an incremental update module; Figure 4 It is a schematic structural diagram of the feature extraction module and the remaining useful life calculation module in the embodiment of the present application. Detailed Embodiments

[0022] To make the objectives, features, and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the embodiments described below are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention. The principles and features of the present invention will be described below with reference to the accompanying drawings, and the examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0023] The term "including" and other similar expressions in the description or claims of the present invention and the above-mentioned drawings mean covering non-exclusive inclusion. For example, a process, method, system, or device that includes a series of steps or units is not limited to the listed steps or units.

[0024] Example 1: As Figure 1 shown, a method for predicting the remaining service life of a lithium battery includes the following steps: S101: Obtain the key parameters of the lithium battery during operation. The key parameters include geographical location parameters, environmental parameters, remaining capacity of the lithium-ion battery, internal temperature of the battery, current and voltage during discharge, number of charge-discharge cycles, and battery health status parameters; The purpose of this step is to collect various parameters closely related to the working state of the lithium battery. Among them, the geographical location parameters include the geographical location information of the lithium battery, such as longitude and latitude, etc., which will affect the working environment of the battery. For example, the climate conditions vary in different regions; The environmental parameters cover environmental factors such as temperature, humidity, and air pressure. These factors have a significant impact on the performance and life of the lithium battery. For example, high temperature may accelerate the chemical reactions inside the battery and shorten the battery life; In this embodiment, the environmental parameters include environmental temperature, humidity, and atmospheric pressure data; the geographical location parameters obtain longitude, latitude coordinates, and altitude values through a GPS module; The remaining capacity of the lithium-ion battery reflects how much electricity the battery can still store currently and is an important indicator for measuring the available energy of the battery; The actual temperature inside the battery, too high or too low temperature will affect the charge-discharge efficiency and safety of the battery; Current and voltage during discharge: When the battery is discharging, monitor the changes in current and voltage in real time. These parameters can reflect the working state and performance of the battery.

[0025] The number of charge-discharge cycles records the number of complete charge-discharge cycles the battery has experienced. Generally speaking, the more the number of cycles, the more the performance of the battery will gradually decline.

[0026] The battery health status parameters are indicators used to evaluate the overall health status of the battery, such as the internal resistance of the battery, the degree of capacity attenuation, etc.; S102: Monitor the key parameters and dynamically adjust the weights of each parameter based on a preset weight adjustment algorithm. The weight adjustment algorithm determines the weight values of each parameter according to the historical change trend of the parameter and its correlation with the remaining service life of the battery; This step continuously monitors the key parameters obtained in S101. Then, the weight adjustment algorithm determines the weight values of each parameter based on the historical change trend of the parameters and their correlation with the remaining service life of the battery. For example, if it is found that the change trend of a certain environmental parameter in the past period has a strong correlation with the shortening of the battery life, the weight of this parameter will increase accordingly; conversely, if a parameter has little relation to the battery life, its weight may decrease. This way of dynamically adjusting the weights can make the subsequent prediction of the remaining service life of the battery more accurate because the influence degrees of different parameters on the battery life are different under different circumstances.

[0027] S103: Construct an input feature matrix for the remaining service life of the lithium-ion battery. The matrix integrates spatio-temporal features and battery state features. This step integrates the various parameters obtained previously to construct an input feature matrix. The matrix integrates spatio-temporal features (such as the spatial features represented by geographical location parameters and the parameter changes related to time, etc.) and battery state features (such as the remaining capacity of the battery, internal temperature, etc.). In this way, the multi-faceted information related to the remaining service life of the battery is comprehensively represented, providing more comprehensive and accurate input data for the subsequent model calculation.

[0028] S104: When the geographical location parameter and / or environmental parameter meet the preset rules, call the environmental preset model matching the target climate region to calculate and output the remaining service life of the lithium battery; otherwise, use the general model to calculate and output the remaining service life of the lithium battery. Among them, the environmental preset model is a machine learning model pre-trained with historical degradation data of the target climate region and can be updated online to adapt to environmental changes.

[0029] The preset rules may mean that when the geographical location or environmental conditions where the battery is located meet a specific range or standard, it is considered that the current environment has certain particularities, and a machine learning model pre-trained with historical degradation data of the specific target climate region needs to be used for prediction. For example, in cold polar regions and hot desert regions, the performance of the battery may be different from that in general environments, so a specially trained model is needed to more accurately predict the remaining life.

[0030] Otherwise, when the preset rules are not met, use the general model to calculate and output the remaining service life of the lithium battery. The general model is applicable to general environmental conditions.

[0031] In addition, the environmental preset model can be updated online to adapt to environmental changes. As time goes by, environmental conditions may change, or new data is continuously generated. By updating the model online, the model can always maintain its adaptability to the current environment, improving the accuracy and reliability of the prediction.

[0032] Exemplarily, the environment preset model: A dedicated model pre-trained for typical climates such as tropical / cold / high-altitude, etc. General model: A benchmark model trained based on global data Switching logic: Determine the current climate region through a decision tree or rule engine (e.g., when the temperature > 40°C and humidity > 80%, activate the tropical model) In a possible implementation, the input feature matrix contains a three-dimensional spatio-temporal tensor associated with a timestamp, where the first dimension is the quantization value of the geographical location coordinates, the second dimension is the parameter change rate within the time window, and the third dimension contains the battery state feature vector, and the feature vector is dimensionally reduced through the principal component analysis method.

[0033] In this implementation, the input feature matrix is the key data structure of the model for predicting the remaining life of lithium batteries. It exists in the form of a three-dimensional spatio-temporal tensor and is associated with the timestamp, which can reflect the chronological order of the data and is conducive to analyzing the change of battery performance over time.

[0034] The first dimension is the quantization value of the geographical location coordinates: The location where the lithium battery is located affects its performance and life. By quantifying geographical location coordinates such as longitude and latitude into numerical values according to rules, the model can consider the impact of the environment in different regions (such as climate differences) on the battery based on this value.

[0035] The second dimension is the parameter change rate within the time window: Monitor key parameters within a specific time period (time window) and calculate their change rates, such as the change rate of battery temperature. The change rate can reflect the dynamic change trend of the parameters and provide useful information for predicting the battery life.

[0036] The third dimension is the battery state feature vector dimensionally reduced by principal component analysis: The battery state feature vector combines multiple parameters such as remaining capacity, temperature, current, and voltage. These parameters were originally high-dimensional and may be redundant. Using the principal component analysis method (PCA) for dimensional reduction can retain key information, remove redundancy, reduce the data dimension, improve the calculation efficiency and performance of the model, and better predict the remaining life of the battery.

[0037] In a possible implementation, the determination of the preset rules uses a climate pattern recognition model under the transfer learning framework, specifically including: inputting real-time geographical location parameters into a pre-trained ResNet-Transformer hybrid network to output the probability distribution of climate regions; when the probability value of the target climate region exceeds the threshold and the environmental parameters match the typical degradation pattern, activate the corresponding environment preset model.

[0038] Among them, transfer learning refers to using the knowledge (pre-trained model) learned on other related tasks or data and applying it to the current task to improve the learning efficiency and performance of the model. The role of the climate pattern recognition model is to identify the climate pattern where the lithium battery is located, so as to provide a basis for subsequent decisions.

[0039] By inputting the real-time geographical location parameters (such as longitude and latitude, etc.) of the lithium battery into a pre-trained hybrid network. After the network's calculation and processing, the output result is the probability distribution regarding different climate regions. For each possible climate region (such as tropical climate region, temperate climate region, frigid climate region, etc.), the network will give the probability value that the current location of the lithium battery belongs to this climate region. For example, the output result may show that the probability that the current location belongs to the tropical climate region is 0.6, the probability that it belongs to the temperate climate region is 0.3, and the probability that it belongs to the frigid climate region is 0.1, etc.

[0040] Exemplarily, assume that a probability threshold is set. When the probability value of a certain target climate region (i.e., the specific climate region of concern) obtained from the above calculation exceeds this threshold, it indicates that the current lithium battery is very likely to be in this target climate region. For example, the threshold is set to 0.5. If the probability value of a certain target climate region is 0.6, this condition is met.

[0041] In addition to considering the climate region probability corresponding to the geographical location, it is also necessary to check whether the current environmental parameters (such as temperature, humidity, air pressure, etc.) match the typical degradation mode of this target climate region. The typical degradation mode refers to the common performance degradation characteristics and rules shown by the lithium battery during long-term use in this climate region. For example, in a hot and humid climate region, the lithium battery may have specific capacity attenuation and internal resistance increase modes. Only when the environmental parameters match the typical degradation mode is it considered that the condition of the preset rule is satisfied.

[0042] When both of the above two conditions are met, that is, the probability value of the target climate region exceeds the threshold and the environmental parameters match the typical degradation mode, the environmental preset model matching this target climate region will be activated. This environmental preset model is a machine learning model pre-trained with historical degradation data for this target climate region before. After activation, it can be used to calculate the remaining service life of the lithium battery more accurately.

[0043] Example 2: As Figure 2 shown, a device for predicting the remaining service life of a lithium battery, the device includes: A parameter acquisition module, used to acquire the key parameters of the lithium battery working, and the key parameters include geographical location parameters, environmental parameters, remaining capacity of the lithium-ion battery, internal battery temperature, current and voltage during discharge, number of charge and discharge cycles, and battery health state parameters; After the parameter acquisition module acquires these key parameters, it outputs and transmits them to the parameter monitoring and weight adjustment module and the feature matrix construction module; The parameter monitoring and weight adjustment module is used to monitor key parameters and dynamically adjust the weights of each parameter based on a preset weight adjustment algorithm. The weight adjustment algorithm determines the weight values of each parameter according to the historical change trend of the parameter and its correlation with the remaining service life of the battery. It receives key parameter data from the parameter acquisition module for monitoring and weight adjustment calculations. After completing the weight adjustment, it is not directly associated with other modules, but the calculated weight values of each parameter will affect the feature matrix construction module because the construction of the feature matrix needs to comprehensively consider each parameter and its weight.

[0044] The feature matrix construction module is used to construct an input feature matrix for the remaining service life of the lithium-ion battery. The matrix integrates spatio-temporal features and battery state features. It obtains key parameter data from the parameter acquisition module and, at the same time, refers to the weight values of each parameter calculated by the parameter monitoring and weight adjustment module, integrates and processes this information, and constructs an input feature matrix. The constructed feature matrix will be passed to the remaining service life calculation module as the input basis for calculating the remaining battery life.

[0045] The remaining service life calculation module, when the geographical location parameter and / or environmental parameter meet the preset rules, calls the environment preset model matching the target climate region to calculate and output the remaining service life of the lithium battery; otherwise, it uses the general model to calculate and output the remaining service life of the lithium battery. Among them, the environment preset model is a machine learning model pre-trained on the historical degradation data of the target climate region and can be updated online to adapt to environmental changes.

[0046] The output module is used to display or transmit the calculated remaining service life of the lithium battery.

[0047] As Figure 3 shown, in a possible implementation manner, the device further includes: The feature extraction module is used to perform learnable PCA dimensionality reduction on the battery operating parameters to generate a spatio-temporal fusion feature matrix. The main function of the feature extraction module is to process the battery operating parameters. Specifically, it uses a learnable principal component analysis (PCA) dimensionality reduction method to generate a spatio-temporal fusion feature matrix. After performing learnable PCA dimensionality reduction on the battery operating parameters, a spatio-temporal fusion feature matrix is generated. The matrix integrates feature information in two dimensions: time and space. For example, the geographical location parameter reflects the spatial feature, and the change of the battery operating parameters over time (such as temperature change and capacity change at different time points) reflects the time feature. Through this fusion, information from multiple aspects is integrated into one matrix, providing a more concise and representative data form for subsequent analysis and prediction.

[0048] The input data of this module comes from the battery operating parameters obtained by the parameter acquisition module. The spatio-temporal fusion feature matrix generated after processing can, on the one hand, provide more refined feature data for the feature matrix construction module to assist it in constructing a more effective input feature matrix; on the other hand, it may also directly participate in the calculation process of the remaining useful life calculation module to provide better input for the prediction of the battery's remaining life.

[0049] Incremental update module, which performs differential parameter update when the environment mutates or the prediction error exceeds the threshold. The incremental update module needs to obtain environmental parameter information (from the parameter acquisition module) to determine whether an environmental mutation has occurred, and also needs to obtain the prediction error information of the remaining useful life calculation module to determine whether the threshold has been exceeded. When performing differential parameter update, it will update the parameters in related modules such as the environmental preset model and the general model, thus affecting the subsequent calculation and prediction processes of these modules.

[0050] Exemplarily, the incremental update module triggers the update process when the following conditions are met simultaneously: The rate of change of environmental temperature ≥ 5°C / hour; The prediction error exceeds 10% of the preset threshold for 3 consecutive times; The updated data transmission adopts incremental compression, and the data integrity is verified through a hash check code. Through the dual judgment of the temperature mutation rate and error persistence, false triggering can be avoided. Adopting incremental compression for updated data transmission can significantly reduce system resource consumption while ensuring update reliability, meeting the deployment requirements of edge computing devices.

[0051] As Figure 4 shown, the feature extraction module further includes: Hash encoding unit, which is used to discretize temperature and cycle times into hash values of a preset length; Coding example: Temperature range | Hash encoding [-20,0)°C|0x5A [0,25)°C|0x3B [25,60]°C|0x8F Instead of floating-point arithmetic, The hash value of the preset length is used as the input of the remaining useful life calculation module to calculate the remaining useful life of the lithium battery.

[0052] The hash encoding unit is used to discretize specific battery operating parameters, namely temperature and number of cycles, and convert them into hash values of a preset length. Discretization divides a continuous numerical range into several discrete intervals, which can simplify the representation and processing of data. Through hash encoding, the originally continuous temperature values and number of cycle values are mapped to specific hash values, and these hash values have a fixed preset length, facilitating subsequent data processing and calculation.

[0053] Under normal circumstances, parameters such as temperature and number of cycles may exist in the form of floating-point numbers in data processing. Floating-point operations are relatively complex and may consume more computing resources and time in some calculation scenarios. By discretizing temperature and number of cycles into hash values and using these hash values to replace the original floating-point numbers for subsequent operations and processing. Hash values are generally fixed-length encodings, which are more efficient and concise in storage and calculation compared to floating-point numbers, can reduce the amount of calculation, and improve the speed and efficiency of data processing.

[0054] The hash values of the preset length obtained after being processed by the hash encoding unit are passed as inputs to the remaining useful life calculation module. The remaining useful life calculation module uses the hash values and combines other relevant input information (such as features obtained after processing other battery operating parameters) to calculate the remaining useful life of the lithium battery.

[0055] The remaining useful life calculation module includes a climate judgment module, and the climate judgment module is used to match the corresponding climate classification code according to the geographical location coordinates; The climate judgment module performs the following operations: Maps the longitude and latitude coordinates to a climate classification code; Matches the preset model identifier through a look-up table method.

[0056] For example, maps the longitude and latitude coordinates to a 3-digit climate classification code, and the coding rules include: The first digit: temperature zone (1 - tropical, 2 - subtropical...); The second digit: humidity characteristic (A - arid, B - humid...); The last digit: season pattern (S - rainy in summer, W - rainy in winter). Finally, matches the preset model identifier through a look-up table method.

[0057] The climate judgment module determines the corresponding climate classification based on the geographical location information of the lithium battery and finds the matching preset model identifier. The remaining useful life calculation module calls the corresponding model according to the preset model identifier provided by the climate judgment module and combines other input information (such as battery operating parameters) to accurately calculate the remaining useful life of the lithium battery.

[0058] The technical means disclosed by the solution of the present invention are not limited to the technical means disclosed in the above embodiments, and also include technical solutions composed of any combination of the above technical features. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A method for predicting the remaining useful life of a lithium battery, characterized in that, It includes the following steps: Obtain the key parameters of the lithium battery operation, where the key parameters include geographical location parameters, environmental parameters, remaining capacity of the lithium-ion battery, internal battery temperature, current and voltage during discharge, number of charge-discharge cycles, and battery health status parameters; Monitor the key parameters and dynamically adjust the weights of each parameter based on a preset weight adjustment algorithm, where the weight adjustment algorithm determines the weight values of each parameter according to the historical change trend of the parameter and its correlation with the remaining service life of the battery; Construct an input feature matrix for the remaining service life of the lithium-ion battery, where the matrix integrates spatio-temporal features and battery state features; When the geographical location parameter and / or environmental parameter meet the preset rules, call the environmental preset model matching the target climate region to calculate and output the remaining service life of the lithium battery; otherwise, use the general model to calculate and output the remaining service life of the lithium battery; where the environmental preset model is a machine learning model pre-trained with historical degradation data of the target climate region and can be updated online to adapt to environmental changes.

2. The method for predicting the remaining service life of a lithium battery according to claim 1, characterized in that, The environmental parameters include environmental temperature, humidity, and atmospheric pressure data; the geographical location parameters are obtained by the GPS module to obtain the longitude and latitude coordinates and altitude value.

3. A method for predicting the remaining service life of a lithium battery according to claim 1, characterized in that, The input feature matrix contains a three-dimensional spatio-temporal tensor associated with a timestamp, where the first dimension is the quantization value of the geographical location coordinates, the second dimension is the parameter change rate within the time window, and the third dimension contains the battery state feature vector, and the feature vector is processed by dimensionality reduction using the principal component analysis method.

4. A method for predicting the remaining service life of a lithium battery according to claim 1, characterized in that, The determination of the preset rules uses a climate pattern recognition model under the transfer learning framework, specifically including: inputting the real-time geographical location parameters into the pre-trained ResNet-Transformer hybrid network to output the climate region probability distribution; when the target climate region probability value exceeds the threshold and the environmental parameters match the typical degradation pattern, activate the corresponding environmental preset model.

5. A device for predicting the remaining service life of a lithium battery, characterized in that, It includes: A parameter acquisition module for obtaining the key parameters of the lithium battery operation, where the key parameters include geographical location parameters, environmental parameters, remaining capacity of the lithium-ion battery, internal battery temperature, current and voltage during discharge, number of charge-discharge cycles, and battery health status parameters; A parameter monitoring and weight adjustment module for monitoring the key parameters and dynamically adjusting the weights of each parameter based on a preset weight adjustment algorithm, where the weight adjustment algorithm determines the weight values of each parameter according to the historical change trend of the parameter and its correlation with the remaining service life of the battery; A feature matrix construction module for constructing an input feature matrix for the remaining service life of the lithium-ion battery, where the matrix integrates spatio-temporal features and battery state features; A remaining service life calculation module, when the geographical location parameter and / or environmental parameter meet the preset rules, call the environmental preset model matching the target climate region to calculate and output the remaining service life of the lithium battery; otherwise, use the general model to calculate and output the remaining service life of the lithium battery; where the environmental preset model is a machine learning model pre-trained with historical degradation data of the target climate region and can be updated online to adapt to environmental changes; An output module for displaying or transmitting the calculated remaining service life of the lithium battery.

6. The device for predicting the remaining service life of a lithium battery according to claim 5, wherein, The device further includes: A feature extraction module for performing learnable PCA dimensionality reduction on battery operating parameters to generate a spatio-temporal fusion feature matrix; An incremental update module for performing differential parameter update when the environment mutates or the prediction error exceeds the threshold.

7. The device for predicting the remaining service life of a lithium battery according to claim 6, wherein, The feature extraction module further includes: A hash coding unit for discretizing temperature and number of cycles into hash values of a preset length; The hash values of the preset length are used as inputs to the remaining service life calculation module to calculate the remaining service life of the lithium battery.

8. The device for predicting the remaining service life of a lithium battery according to claim 6, characterized in that, The remaining service life calculation module includes a climate judgment module for matching corresponding climate classification codes according to geographical location coordinates; The climate judgment module performs the following operations: Mapping the longitude and latitude coordinates to climate classification codes; Matching a preset model identifier through a look-up table method.

9. The device for predicting the remaining service life of a lithium battery according to claim 6, characterized in that: The incremental update module triggers an update process when the following conditions are simultaneously met: The environmental temperature change rate ≥ 5 °C / hour; The prediction error exceeds 10% of the preset threshold for 3 consecutive times; The updated data transmission adopts an incremental compression method, and the data integrity is verified through a hash check code.

Images