New energy automobile battery health prediction method and system

Abstract

The invention discloses a new energy automobile battery health prediction method and system, and relates to the technical field of monitoring analysis, and the method comprises the steps: carrying out the real-time collection of the operation state parameters of a power battery pack corresponding to a new energy automobile, obtaining a battery state data set corresponding to the new energy automobile, carrying out the adaptive noise suppression processing of the battery state data set, and obtaining a new energy automobile battery health prediction result. Generating a de-noised battery state feature map, performing feature extraction on the de-noised battery state feature map to obtain a multi-dimensional coupling feature set, constructing a dynamic weight distribution matrix, performing feature fusion on the multi-dimensional coupling feature set, determining a battery health state comprehensive index, and triggering a grading early warning mechanism based on the battery health state comprehensive index; and constructing a battery health prediction model, carrying out long-term modeling on the battery health state comprehensive index, predicting a battery health state evolution trend in a future charging and discharging period, generating a balance maintenance strategy, and pushing the balance maintenance strategy to the vehicle-mounted terminal and the cloud management platform. The method has the effect of improving the accuracy of battery health prediction.

Description

Technical Field

[0001] This application relates to the technical field of monitoring and analysis, and in particular to a method and system for predicting the health of new energy vehicle batteries. Background Art

[0002] The lifespan issue of power batteries is a key problem for new energy vehicles. In the field of power batteries, it is generally believed that when the battery capacity drops to 80% of the rated capacity, it can no longer be used in electric vehicles. There are certain correlations between the internal characteristics of single cells, the grouping method of power sources, and the performance requirements of the whole vehicle and the lifespan state of the battery. The so-called calendar life of the battery refers to the period from the production date of the battery to the expiration date, including the impacts of factors such as working conditions, temperature, cycling, shelving, and aging on the battery life. The end of the battery life of one battery in a system often affects the operation of the entire system, resulting in the overall failure of the system's functions. The accurate prediction of the battery health state is not only related to the current usage of the battery but also related to the recycling and cascaded utilization of old batteries. Therefore, the correct estimation of the battery health state and the research on the calendar life can further guide the operation of the battery, provide data support for the establishment of the battery health management system, and be of great significance for extending the service life of power batteries.

[0003] In related technologies, in the existing methods for predicting the health of new energy vehicle batteries, it is impossible to analyze multiple states of the battery and the continuous operation data of its entire life cycle, and the accuracy and reliability of battery health prediction are relatively low, and there are areas for improvement. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, this application provides a method and system for predicting the health of new energy vehicle batteries.

[0005] In a first aspect, this application provides a method for predicting the health of new energy vehicle batteries, including the following steps: Step S1: Real-time collect the operation state parameters of the power battery pack corresponding to the new energy vehicle, and then obtain the battery state data set corresponding to the new energy vehicle. The battery state data set includes voltage fluctuation time series data, current ripple spectrum data, and temperature gradient distribution data; Step S2: Perform adaptive noise suppression processing on the battery state data set, generate a denoised battery state feature map based on the result of the noise suppression processing, and extract features from the denoised battery state feature map, and then obtain a multi-dimensional coupling feature set. The multi-dimensional coupling feature set includes capacity decay trajectory features, internal resistance mutation trend features, and polarization effect accumulation features; Step S3: Construct a dynamic weight assignment matrix, perform feature fusion on the multi-dimensional coupled feature set based on the dynamic weight assignment matrix, then confirm the comprehensive battery health state index based on the result of the feature fusion, and trigger a hierarchical early warning mechanism based on the comprehensive battery health state index; Step S4: Construct a battery health prediction model, perform long-term modeling on the comprehensive battery health state index, and predict the evolution trend of the battery health state in future charge and discharge cycles; Step S5: Generate an equalization maintenance strategy according to the evolution trend of the battery health state in future charge and discharge cycles and push it to the in-vehicle terminal and the cloud management platform.

[0006] Preferably, the specific steps of the step S1 are as follows: Step S11: Collect the voltage waveform changes through the voltage sensor array installed inside the power battery pack, and then construct the voltage fluctuation time series data based on the voltage waveform changes; Step S12: Use a high-frequency current sensor to capture the AC ripple signal during the charge and discharge process, and generate the current ripple spectrum data through Fourier transform; Step S13: Arrange a three-dimensional temperature sensing network inside the battery pack, and then record the temperature gradient distribution data based on the three-dimensional temperature sensing network; Step S14: Align the voltage fluctuation time series data, the current ripple spectrum data, and the temperature gradient distribution data, and form a battery state data set based on the result of the alignment process.

[0007] Preferably, the specific steps of the step S2 are as follows: Step S21: Perform sliding window noise separation on the voltage fluctuation time series data, and then separate the voltage fluctuation features; Step S22: Perform spectral line density compensation on the current ripple spectrum data, and then reconstruct the current frequency domain features based on the result of the spectral line density compensation; Step S23: Denoise the temperature gradient distribution data, and then obtain the temperature distribution features according to the result of the denoising process; Step S24: Perform time-frequency domain fusion on the voltage fluctuation features, the current frequency domain features, and the temperature distribution features to generate a denoised battery state feature map; Step S25: Perform hierarchical feature extraction on the denoised battery state feature map through a deep convolutional neural network, and then layer by layer separate a multi-dimensional coupled feature set, where the multi-dimensional coupled feature set includes capacity attenuation trajectory features, internal resistance mutation trend features, and polarization effect accumulation features.

[0008] Preferably, the specific steps of the step S3 are as follows: Step S31: Construct a dynamic weight allocation matrix based on the cyclic charge and discharge historical data of the power battery pack of a new energy vehicle, and then dynamically allocate the weight coefficients corresponding to the capacity attenuation trajectory feature, the internal resistance mutation trend feature, and the polarization effect accumulation feature based on the dynamic weight allocation matrix; Step S32: Normalize the capacity attenuation trajectory feature, the internal resistance mutation trend feature, and the polarization effect accumulation feature corresponding to the multi-dimensional coupling feature set, and then confirm the standardized indices corresponding to the capacity attenuation trajectory feature, the internal resistance mutation trend feature, and the polarization effect accumulation feature; Step S33: Perform comprehensive weighted calculation according to the weight coefficients corresponding to the capacity attenuation trajectory feature, the internal resistance mutation trend feature, and the polarization effect accumulation feature and the standardized indices corresponding to the capacity attenuation trajectory feature, the internal resistance mutation trend feature, and the polarization effect accumulation feature to generate a comprehensive battery health state index.

[0009] Preferably, after generating the comprehensive battery health state index, it specifically further includes: Perform a health risk assessment on the power battery pack corresponding to the new energy vehicle based on the comprehensive battery health state index. The process of the health risk assessment is as follows: Compare the comprehensive battery health state index with a preset battery health state threshold. The preset battery health state threshold includes a first battery health state threshold and a second battery health state threshold, where the first battery health state threshold is less than the second battery health state threshold; If the comprehensive battery health state index exceeds the preset second battery health state threshold, no warning is required; If the comprehensive battery health state index is between the preset first battery health state threshold and the preset second battery health state threshold, a first-level warning is given, and the power battery pack corresponding to the new energy vehicle is adjusted according to a preset first adjustment method; If the comprehensive battery health state index is lower than the preset first battery health state threshold, a second-level warning is given, and the power battery pack corresponding to the new energy vehicle is adjusted according to a preset second adjustment method.

[0010] Preferably, step S4 specifically includes: Within a preset time period, collect the real-time comprehensive battery health state index corresponding to the power battery pack of the new energy vehicle to form a time series, and then obtain a curve graph of the change of the real-time comprehensive battery health state index with the time series ; Compare the curve graph of the change of the real-time comprehensive battery health state index with the time series with the curve graph of the change of the standard comprehensive battery health state index, and then construct a battery health prediction model; Predict the evolution trend of the battery health state in the future charge and discharge cycles according to the battery health prediction model, and then confirm the change coefficient corresponding to the comprehensive battery health state index based on the evolution trend of the battery health state in the future charge and discharge cycles.

[0011] Preferably, after confirming the change coefficient corresponding to the comprehensive battery health state index, it further includes: Compare the change coefficient corresponding to the comprehensive battery health state index with a preset change threshold; If the change coefficient corresponding to the comprehensive battery health state index exceeds the preset change threshold, generate an equalization maintenance strategy and push it to the in-vehicle terminal and the cloud management platform; otherwise, there is no need to generate an equalization maintenance strategy.

[0012] In a second aspect, the present application provides a new energy vehicle battery health prediction system, including: A data acquisition module, configured to collect the operating state parameters of the power battery pack corresponding to the new energy vehicle in real time, and then obtain the battery state data set corresponding to the new energy vehicle. The battery state data set includes voltage fluctuation time series data, current ripple spectrum data, and temperature gradient distribution data; A feature extraction module, configured to perform adaptive noise suppression processing on the battery state data set, generate a denoised battery state feature map based on the result of the noise suppression processing, and extract features from the denoised battery state feature map, and then obtain a multi-dimensional coupling feature set. The multi-dimensional coupling feature set includes capacity attenuation trajectory features, internal resistance mutation trend features, and polarization effect accumulation features; A confirmation module, configured to construct a dynamic weight allocation matrix, perform feature fusion on the multi-dimensional coupling feature set based on the dynamic weight allocation matrix, and then confirm the comprehensive battery health state index based on the result of the feature fusion, and trigger a hierarchical warning mechanism based on the comprehensive battery health state index; A prediction module, configured to construct a battery health prediction model, perform long-term modeling on the comprehensive battery health state index, and predict the evolution trend of the battery health state in the future charge and discharge cycles; A maintenance module, configured to generate an equalization maintenance strategy according to the evolution trend of the battery health state in the future charge and discharge cycles and push it to the in-vehicle terminal and the cloud management platform.

[0013] In a third aspect, the present application provides a computer-readable storage medium storing instructions, which when run on a computer, cause the computer to execute a new energy vehicle battery health prediction method as described in any one of the above.

[0014] In summary, the present application includes the following beneficial technical effects: The present application provides a method for predicting the health of a new energy vehicle battery. By performing adaptive noise suppression processing and feature extraction on the battery state data set corresponding to the new energy vehicle, a multi-dimensional coupled feature set is obtained, and a comprehensive battery health state index is confirmed based on the multi-dimensional coupled feature set. Furthermore, a hierarchical warning mechanism is triggered based on the comprehensive battery health state index, thereby effectively managing the battery health, reducing the occurrence of situations where the driving of the new energy vehicle is affected due to battery health problems, and performing long-term modeling on the comprehensive battery health state index to predict the evolution trend of the battery health state in future charge and discharge cycles, and generating an equalization maintenance strategy and pushing it to the in-vehicle terminal and the cloud management platform, thereby effectively predicting the battery health, and thus effectively improving the accuracy and reliability of the battery health prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0016] Figure 1 is a flowchart of the method for predicting the health of a new energy vehicle battery according to an embodiment of the present application.

[0017] Figure 2 is a schematic diagram of the system for predicting the health of a new energy vehicle battery according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The following will further describe the present application in detail Figure 1-2 with reference to the appended

[0019] Embodiment 1: An embodiment of the present application discloses a method for predicting the health of a new energy vehicle battery.

[0020] Referring to Figure 1 , a method for predicting the health of a new energy vehicle battery includes the following steps: Step S1: Real-time collect the operating state parameters of the power battery pack corresponding to the new energy vehicle, and thus obtain the battery state data set corresponding to the new energy vehicle. The battery state data set includes voltage fluctuation time series data, current ripple spectrum data, and temperature gradient distribution data; Step S2: Perform adaptive noise suppression processing on the battery state data set, generate a denoised battery state feature map based on the result of the noise suppression processing, and perform feature extraction on the denoised battery state feature map, thereby obtaining a multi-dimensional coupled feature set. The multi-dimensional coupled feature set includes capacity attenuation trajectory features, internal resistance mutation trend features, and polarization effect accumulation features; Step S3: Construct a dynamic weight allocation matrix, perform feature fusion on the multi-dimensional coupled feature set based on the dynamic weight allocation matrix, then confirm the comprehensive battery health state index based on the result of the feature fusion, and trigger a hierarchical warning mechanism based on the comprehensive battery health state index; Step S4: Construct a battery health prediction model, perform long-term modeling on the comprehensive battery health state index, and predict the evolution trend of the battery health state in future charge and discharge cycles; Step S5: Generate an equalization maintenance strategy according to the evolution trend of the battery health state in future charge and discharge cycles and push it to the in-vehicle terminal and the cloud management platform.

[0021] It should be noted that the specific steps of step S1 are as follows: Step S11: Collect the voltage waveform changes through the voltage sensor array installed inside the power battery pack, and then construct the voltage fluctuation time series data based on the voltage waveform changes; Step S12: Use a high-frequency current sensor to capture the AC ripple signal during the charge and discharge process, and generate the current ripple spectrum data through Fourier transform; Step S13: Arrange a three-dimensional temperature sensing network inside the battery pack, and then record the temperature gradient distribution data based on the three-dimensional temperature sensing network; Step S14: Align the voltage fluctuation time series data, the current ripple spectrum data, and the temperature gradient distribution data, and form a battery state data set based on the result of the alignment process.

[0022] It should be noted that the specific steps of step S2 are as follows: Step S21: Perform sliding window noise separation on the voltage fluctuation time series data, and then separate the voltage fluctuation features; Step S22: Perform spectral line density compensation on the current ripple spectrum data, and then reconstruct the current frequency domain features based on the result of the spectral line density compensation; Step S23: Perform denoising processing on the temperature gradient distribution data, and then obtain the temperature distribution features according to the result of the denoising processing; Step S24: Perform time-frequency domain fusion on the voltage fluctuation features, the current frequency domain features, and the temperature distribution features to generate a denoised battery state feature map; Step S25: Perform hierarchical feature extraction on the denoised battery state feature map through a deep convolutional neural network, and then layer by layer separate the multi-dimensional coupled feature set, and the multi-dimensional coupled feature set includes the capacity attenuation trajectory feature, the internal resistance mutation trend feature, and the polarization effect accumulation feature.

[0023] Specifically, for the voltage fluctuation time series data, a sliding window noise separation technique is adopted. By setting a 10-ms sliding window and using a combined algorithm of median filtering and mean filtering, the high-frequency noise is separated from the real voltage fluctuation characteristics. For example, the voltage spike signal captured during fast charging can accurately extract the voltage change slope of the charge and discharge platform after separation. For the current ripple spectrum data, spectral line density compensation is performed. For the spectral data in the frequency band of 20 Hz - 2000 Hz, the missing spectral lines caused by the sampling interval are filled through an interpolation algorithm, and then the current frequency domain characteristics are reconstructed based on the compensated spectral lines, and the distribution rules of high-frequency harmonic components and low-frequency ripples can be clearly identified. For the temperature gradient distribution data, first, the random noise is removed by Gaussian filtering, and then the heat conduction equation is used to perform spatio-temporal smoothing on the temperature field. For example, when a local overheat occurs in the battery module, the heat diffusion rate of the temperature anomaly area can be accurately extracted. Furthermore, the voltage fluctuation characteristics, current frequency domain characteristics, and temperature distribution characteristics are mapped into the time-frequency domain coordinate system, and after alignment along the time axis, tensor fusion is performed to generate a denoised battery state feature map containing the voltage-current-temperature coupling relationship. Finally, through a deep convolutional neural network, hierarchical extraction is performed on the denoised battery state feature map. The shallow network captures local features such as voltage ripple burrs, and the deep network extracts global features such as the slope of the capacity decay curve, and the capacity decay trajectory features, internal resistance mutation trend features, and polarization effect accumulation features are separated layer by layer. The advantage of the above process is that through multi-dimensional data denoising and fusion, the accuracy of feature extraction is effectively improved. For example, the sliding window technique can increase the signal-to-noise ratio of voltage features by more than 30%, and the spectral line density compensation can reduce the recognition error of current frequency domain features to less than 5%. The hierarchical extraction mechanism of the deep convolutional network can accurately capture the subtle changes in the battery health state and provide a more reliable feature input for subsequent health prediction.

[0024] It should be noted that the step S3 specifically includes the following steps: Step S31: Construct a dynamic weight allocation matrix based on the cyclic charge and discharge historical data of the power battery pack of new energy vehicles, and then dynamically allocate the weight coefficients corresponding to the capacity decay trajectory features, internal resistance mutation trend features, and polarization effect accumulation features based on the dynamic weight allocation matrix; Step S32: Perform normalization processing on the capacity decay trajectory features, internal resistance mutation trend features, and polarization effect accumulation features corresponding to the multi-dimensional coupling feature set, and then confirm the standardized indexes corresponding to the capacity decay trajectory features, internal resistance mutation trend features, and polarization effect accumulation features; Step S33: Perform comprehensive weighted calculation according to the weight coefficients corresponding to the capacity decay trajectory features, internal resistance mutation trend features, and polarization effect accumulation features and the standardized indexes corresponding to the capacity decay trajectory features, internal resistance mutation trend features, and polarization effect accumulation features to generate a comprehensive battery health state index.

[0025] Specifically, a dynamic weight assignment matrix is constructed based on the cyclic charge and discharge historical data of the power battery pack. Specifically, battery health data under different cycle numbers (such as 50 times, 100 times, 200 times) and different working conditions (such as urban roads, highways, low-temperature environments) are collected, and a weight prediction model is trained using the XGBoost algorithm. Exemplarily, when the battery cycle number reaches 150 times and the working temperature is lower than 0 °C, the above model will automatically increase the weight coefficient of the internal resistance mutation trend feature to 0.4 (the initial weight is 0.3) to highlight the impact of internal resistance change on battery health in a low-temperature environment. Then, normalization processing is performed on the capacity attenuation trajectory feature, internal resistance mutation trend feature, and polarization effect accumulation feature in the multi-dimensional coupling feature set, and each feature value is mapped to the [0, 1] interval. Taking the capacity attenuation trajectory feature as an example, if the current capacity retention rate of a certain battery is 85%, the normalization index is 0.85; in the internal resistance mutation trend feature, if the internal resistance increases by 15% compared to the initial value, the normalization index is 0.15 (using reverse mapping, the more it increases, the lower the index). Finally, comprehensive weighted calculation is performed according to the weight coefficient and normalization index of each feature. For example, the comprehensive battery health status index = capacity attenuation trajectory feature normalization index × 0.35 + internal resistance mutation trend feature normalization index × 0.4 + polarization effect accumulation feature normalization index × 0.25 (the weight coefficient is dynamically adjusted according to the working condition).

[0026] Further, after generating the comprehensive battery health status index, it specifically further includes: Performing a health risk assessment on the power battery pack corresponding to the new energy vehicle based on the comprehensive battery health status index, and the process of the health risk assessment is as follows: Comparing the comprehensive battery health status index with a preset battery health status threshold, and the preset battery health status threshold includes a first battery health status threshold and a second battery health status threshold, where the first battery health status threshold is less than the second battery health status threshold; If the comprehensive battery health status index exceeds the preset second battery health status threshold, no warning is required; If the comprehensive battery health status index is between the preset first battery health status threshold and the preset second battery health status threshold, a first-level warning is issued, and the power battery pack corresponding to the new energy vehicle is adjusted according to a preset first adjustment method; If the comprehensive battery health status index is lower than the preset first battery health status threshold, a second-level warning is issued, and the power battery pack corresponding to the new energy vehicle is adjusted according to a preset second adjustment method.

[0027] Specifically, in the embodiments of the present application, when performing a health risk assessment, it is first necessary to preset battery health state thresholds. For example, the first battery health state threshold is set to 0.8, and the second battery health state threshold is set to 0.9. Then, compare the calculated comprehensive battery health state index with the above two thresholds. If the comprehensive index exceeds 0.9, it indicates that the battery health state is good and there is no need to trigger the warning mechanism. If the comprehensive index is between 0.8 and 0.9, it is determined as a first-level warning. At this time, the first adjustment method is started. Exemplarily, the charging cut-off voltage is lowered from 4.2V to 4.15V, and at the same time, a yellow warning icon is displayed on the in-vehicle terminal, and a prompt message "It is recommended to optimize the charging frequency and avoid full charging" is pushed to the user. If the comprehensive index is lower than 0.8, it is determined as a second-level warning, and the second adjustment method is immediately executed, such as restricting the fast charging power to 30kW, automatically starting the battery balancing program, and generating a forced maintenance work order through the cloud management platform, prompting the user that "the battery health has fallen below the threshold and it is necessary to go to the service station for inspection within 72 hours".

[0028] It should be noted that the specific steps of step S4 include: Within a preset time period, collect the real-time comprehensive battery health state index corresponding to the power battery pack of the new energy vehicle to form a time series, and then obtain a curve graph of the change of the real-time comprehensive battery health state index with the time series ; Based on the comparison between the curve graph of the change of the real-time comprehensive battery health state index with the time series and the standard curve graph of the change of the comprehensive battery health state index, a battery health prediction model is constructed; According to the battery health prediction model, predict the evolution trend of the battery health state in the future charge and discharge cycles, and then confirm the change coefficient corresponding to the comprehensive battery health state index based on the evolution trend of the battery health state in the future charge and discharge cycles.

[0029] Specifically, the battery health prediction model is: ; Among them, represents the preset time period, represents the standard curve graph of the change of the comprehensive battery health state index, represents the change coefficient corresponding to the comprehensive battery health state index; Input the curve graph of the change of the real-time comprehensive battery health state index with the time series into the battery health prediction model, and then confirm the change coefficient corresponding to the comprehensive battery health state index .

[0030] Furthermore, after confirming the change coefficient corresponding to the comprehensive battery health state index, it also includes: Compare the change coefficient corresponding to the comprehensive battery health state index with a preset change threshold; If the change coefficient corresponding to the comprehensive battery health state index exceeds the preset change threshold, an equalization maintenance strategy is generated and pushed to the vehicle-mounted terminal and the cloud management platform. Otherwise, there is no need to generate an equalization maintenance strategy. The equalization maintenance strategy includes, but is not limited to, an equalization charging strategy, a temperature management plan, and suggestions for limiting charge and discharge currents.

[0031] Embodiment 2: This embodiment of the present application also discloses a new energy vehicle battery health prediction system.

[0032] Refer to Figure 2 , a new energy vehicle battery health prediction system, including: A data acquisition module for real-time acquisition of the operating state parameters of the power battery pack corresponding to the new energy vehicle, and thus obtaining the battery state data set corresponding to the new energy vehicle. The battery state data set includes voltage fluctuation time series data, current ripple spectrum data, and temperature gradient distribution data; A feature extraction module for performing adaptive noise suppression processing on the battery state data set, generating a denoised battery state feature map based on the result of the noise suppression processing, and performing feature extraction on the denoised battery state feature map, and thus obtaining a multi-dimensional coupling feature set. The multi-dimensional coupling feature set includes a capacity attenuation trajectory feature, an internal resistance mutation trend feature, and a polarization effect accumulation feature; A confirmation module for constructing a dynamic weight allocation matrix, performing feature fusion on the multi-dimensional coupling feature set based on the dynamic weight allocation matrix, and thus confirming the comprehensive battery health state index based on the result of the feature fusion, and triggering a hierarchical warning mechanism based on the comprehensive battery health state index; A prediction module for constructing a battery health prediction model, performing long-term modeling on the comprehensive battery health state index, and predicting the evolution trend of the battery health state in future charge and discharge cycles; A maintenance module for generating an equalization maintenance strategy according to the evolution trend of the battery health state in future charge and discharge cycles and pushing it to the vehicle-mounted terminal and the cloud management platform.

[0033] The above content is only an example and explanation of the concept of the present invention. Those skilled in the art of the present technology make various modifications or supplements to the described specific embodiments or use similar methods to replace them. As long as they do not deviate from the concept of the invention, they should all fall within the protection scope of the present invention.

[0034] In the description of this specification, the descriptions referring to terms such as "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in a suitable manner in any one or more embodiments or examples.

[0035] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only the specific implementation manners. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification in order to better explain the principles and practical applications of the present invention, so that those skilled in the relevant technical fields can well understand and utilize the present invention.

Claims

1. A method for predicting the health of a new energy vehicle battery, characterized in that, Including the following steps: Step S1: Collect the real-time operating state parameters of the power battery pack corresponding to the new energy vehicle, and then obtain the battery state data set corresponding to the new energy vehicle. The battery state data set includes voltage fluctuation time series data, current ripple spectrum data, and temperature gradient distribution data; Step S2: Perform adaptive noise suppression processing on the battery state data set, generate a denoised battery state feature map based on the result of the noise suppression processing, and perform feature extraction on the denoised battery state feature map, and then obtain a multi-dimensional coupling feature set. The multi-dimensional coupling feature set includes capacity attenuation trajectory features, internal resistance mutation trend features, and polarization effect accumulation features; Step S3: Construct a dynamic weight allocation matrix, perform feature fusion on the multi-dimensional coupling feature set based on the dynamic weight allocation matrix, and then confirm the comprehensive battery health state index based on the result of the feature fusion, and trigger a hierarchical warning mechanism based on the comprehensive battery health state index; Step S4: Construct a battery health prediction model, perform long-term modeling on the comprehensive battery health state index, and predict the evolution trend of the battery health state in future charge and discharge cycles; Step S5: Generate an equalization maintenance strategy according to the evolution trend of the battery health state in future charge and discharge cycles and push it to the vehicle-mounted terminal and the cloud management platform.

2. The method for predicting the health of a new energy vehicle battery according to claim 1, wherein, The specific steps of step S1 include the following steps: Step S11: Collect the voltage waveform changes through the voltage sensor array installed inside the power battery pack, and then construct the voltage fluctuation time series data based on the voltage waveform changes; Step S12: Use a high-frequency current sensor to capture the AC ripple signal during the charge and discharge process, and generate the current ripple spectrum data through Fourier transform; Step S13: Arrange a three-dimensional temperature sensing network inside the battery pack, and then record the temperature gradient distribution data based on the three-dimensional temperature sensing network; Step S14: Align the voltage fluctuation time series data, current ripple spectrum data, and temperature gradient distribution data, and form a battery state data set based on the result of the alignment process.

3. A method for predicting the health of a new energy vehicle battery according to claim 1, characterized in that, The specific steps of step S2 include the following steps: Step S21: Perform sliding window noise separation on the voltage fluctuation time series data, and then separate the voltage fluctuation features; Step S22: Perform spectral line density compensation on the current ripple spectrum data, and then reconstruct the current frequency domain features based on the result of the spectral line density compensation; Step S23: Perform denoising processing on the temperature gradient distribution data, and then obtain the temperature distribution features according to the result of the denoising processing; Step S24: Perform time-frequency domain fusion on the voltage fluctuation features, current frequency domain features, and temperature distribution features to generate a denoised battery state feature map; Step S25: Perform hierarchical feature extraction on the denoised battery state feature map through a deep convolutional neural network, and then gradually separate the multi-dimensional coupling feature set. The multi-dimensional coupling feature set includes capacity attenuation trajectory features, internal resistance mutation trend features, and polarization effect accumulation features.

4. A method for predicting the health of a new energy vehicle battery according to claim 1, characterized in that, The specific steps of step S3 include the following steps: Step S31: Construct a dynamic weight allocation matrix based on the cyclic charge and discharge historical data of the power battery pack of a new energy vehicle, and then dynamically allocate the weight coefficients corresponding to the capacity attenuation trajectory feature, the internal resistance mutation trend feature, and the polarization effect accumulation feature based on the dynamic weight allocation matrix; Step S32: Normalize the capacity attenuation trajectory feature, the internal resistance mutation trend feature, and the polarization effect accumulation feature corresponding to the multi-dimensional coupling feature set, and then confirm the standardized indexes corresponding to the capacity attenuation trajectory feature, the internal resistance mutation trend feature, and the polarization effect accumulation feature; Step S33: Perform comprehensive weighted calculation according to the weight coefficients corresponding to the capacity attenuation trajectory feature, the internal resistance mutation trend feature, and the polarization effect accumulation feature and the standardized indexes corresponding to the capacity attenuation trajectory feature, the internal resistance mutation trend feature, and the polarization effect accumulation feature to generate a comprehensive battery health state index.

5. A method for predicting the health of a new energy vehicle battery according to claim 1, characterized in that, After generating the comprehensive battery health state index, it specifically further includes: Perform a health risk assessment on the power battery pack corresponding to the new energy vehicle based on the comprehensive battery health state index. The process of the health risk assessment is as follows: Compare the comprehensive battery health state index with the preset battery health state thresholds. The preset battery health state thresholds include a first battery health state threshold and a second battery health state threshold, where the first battery health state threshold is less than the second battery health state threshold; If the comprehensive battery health state index exceeds the preset second battery health state threshold, no warning is required; If the comprehensive battery health state index is between the preset first battery health state threshold and the preset second battery health state threshold, a first-level warning is issued, and the power battery pack corresponding to the new energy vehicle is adjusted according to the preset first adjustment method; If the comprehensive battery health state index is lower than the preset first battery health state threshold, a second-level warning is issued, and the power battery pack corresponding to the new energy vehicle is adjusted according to the preset second adjustment method.

6. The method for predicting the health of a new energy vehicle battery according to claim 1, wherein, The specific content of step S4 includes: Within a preset time period, collect the real-time comprehensive battery health state index corresponding to the power battery pack of a new energy vehicle to form a time series, and then obtain a curve graph showing the change of the real-time comprehensive battery health state index with the time series ; Compare the curve graph of the real-time comprehensive battery health state index changing with the time series with the curve graph of the standard comprehensive battery health state index changing, and then construct a battery health prediction model; Predict the evolution trend of the battery health state in the future charge and discharge cycles according to the battery health prediction model, and then confirm the change coefficient corresponding to the comprehensive battery health state index based on the evolution trend of the battery health state in the future charge and discharge cycles.

7. A method for predicting the health of a new energy vehicle battery according to claim 6, characterized in that After confirming the change coefficient corresponding to the comprehensive battery health state index, it further includes: Compare the change coefficient corresponding to the comprehensive battery health state index with the preset change threshold; If the change coefficient corresponding to the comprehensive battery health state index exceeds the preset change threshold, an equalization maintenance strategy is generated and pushed to the in-vehicle terminal and the cloud management platform. Otherwise, no equalization maintenance strategy needs to be generated.

8. A new energy vehicle battery health prediction system, which is applied to the new energy vehicle battery health prediction method described in any one of the above claims 1-7, is characterized in that, It includes: A data acquisition module for real-time acquisition of the operating state parameters of the power battery pack corresponding to a new energy vehicle, so as to obtain a battery state data set corresponding to the new energy vehicle, where the battery state data set includes voltage fluctuation time series data, current ripple spectrum data, and temperature gradient distribution data; A feature extraction module for performing adaptive noise suppression processing on the battery state data set, generating a denoised battery state feature map based on the result of the noise suppression processing, and extracting features from the denoised battery state feature map, so as to obtain a multi-dimensional coupling feature set, where the multi-dimensional coupling feature set includes capacity attenuation trajectory features, internal resistance mutation trend features, and polarization effect accumulation features; A confirmation module for constructing a dynamic weight allocation matrix, performing feature fusion on the multi-dimensional coupling feature set based on the dynamic weight allocation matrix, and then confirming a comprehensive battery health state index based on the result of the feature fusion, and triggering a hierarchical early warning mechanism based on the comprehensive battery health state index; A prediction module for constructing a battery health prediction model, performing long-term modeling on the comprehensive battery health state index, and predicting the evolution trend of the battery health state in future charge and discharge cycles; A maintenance module for generating an equalization maintenance strategy according to the evolution trend of the battery health state in future charge and discharge cycles and pushing it to the in-vehicle terminal and the cloud management platform.

9. A computer-readable storage medium, characterized in that: Stored with instructions that, when the instructions are run on a computer, cause the computer to execute a method for predicting the health of a new energy vehicle battery according to any one of claims 1 to 7.

Images