New energy automobile battery life prediction method and system

Abstract

The application is suitable for the technical field of battery life prediction, and particularly relates to a new energy automobile battery life prediction method and system. The method comprises the following steps: identifying a target object that needs to be subjected to new energy automobile battery life prediction, determining a loading platform of the target object, reading attribute data of the target object, wherein the attribute data at least comprises battery code, cycle number and remaining service life, constructing a use database of the same brand of batteries, selecting a plurality of experience samples from the use database via the cycle number, wherein the cycle number of the experience samples is greater than that of the target object, and configuring influence factors of life prediction. Through calculation of similarity and weight, the experience samples are used to calculate the remaining service life of the target object, effectively solving the problem of insufficient early data of new energy automobile batteries, and greatly improving the prediction accuracy.

Description

Technical Field

[0001] This invention relates to the field of battery life prediction technology, and in particular to a method and system for predicting the battery life of new energy vehicles. Background Technology

[0002] Battery lifespan is a key indicator for evaluating the performance and economy of electric vehicles, and it is typically influenced by multiple factors, including battery type, usage environment, charging method, and usage frequency. Currently, the mainstream power batteries are ternary lithium batteries and lithium iron phosphate batteries. The former has higher energy density but a relatively shorter cycle life, while the latter has a longer cycle life and higher safety. Under ideal conditions, lithium iron phosphate batteries can achieve a cycle life of over 3000 cycles and a service life exceeding 8 to 10 years, while ternary lithium batteries have a lifespan of around 1500 to 2000 cycles. In actual use, frequent fast charging, overcharging and over-discharging, and high or low temperature environments all accelerate battery degradation.

[0003] In existing technologies, a model is typically trained using a large amount of historical battery charge-discharge cycle data to establish a mapping relationship between battery state and lifespan. This model is then used for lifespan prediction. However, in the early stages of battery use, due to insufficient historical usage data, traditional data-driven lifespan prediction models struggle to accurately estimate the battery's health status and remaining lifespan, significantly impacting the accuracy of the prediction.

[0004] Therefore, "how to utilize existing similar batteries to provide early-stage operational data and extrapolate battery life" is the technical problem that this invention aims to solve. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for predicting the lifespan of new energy vehicle batteries, in order to solve the problem mentioned in the background art of "how to use similar batteries to provide early-stage operating data and calculate the battery lifespan".

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A method for predicting the lifespan of new energy vehicle batteries, the method comprising:

[0008] The target object for predicting the battery life of new energy vehicles is identified, the loading platform of the target object is determined, and the attribute data of the target object is read. The attribute data includes at least the battery code, cycle count and remaining service life. A usage database of batteries of the same brand is constructed. Based on the cycle count, several experience samples are selected from the usage database, wherein the cycle count of the experience samples is greater than that of the target object.

[0009] The factors influencing lifespan prediction are configured, including at least temperature, charge / discharge rate and vehicle driving environment. An anomaly detection algorithm is used to clean the experience samples, a similarity calculation formula is constructed, and the similarity between the target object and each experience sample is calculated sequentially.

[0010] By integrating the preset weight formula and the similarity, the weight of each experience sample is calculated. From the usage database, the expected remaining lifespan of each experience sample is traversed. Based on the weight and the expected remaining lifespan, the prediction result of the target object is calculated and sent to the preset terminal.

[0011] Furthermore, the steps of identifying the target object for which new energy vehicle battery life prediction needs to be performed, determining the loading platform of the target object, and reading the attribute data of the target object include:

[0012] A degradation model is created by collecting historical operating data of the loading platform and training the degradation model.

[0013] A time window is constructed, real-time operating condition data within the time window is collected, the real-time operating condition data and the prediction results are input into the degradation model, and the correction value is output to correct the prediction results.

[0014] Furthermore, the step of selecting several experience samples from the database includes:

[0015] Each attribute data is divided into several levels, where the target object and the experience sample correspond to the same level;

[0016] The number of experience samples is counted. When the number is less than a threshold, the level is adjusted, and the adjustment includes at least merging and re-segmenting.

[0017] Furthermore, the steps of cleaning the experience samples using an anomaly detection algorithm, constructing a similarity calculation formula, and sequentially calculating the similarity between the target object and each experience sample include:

[0018] Each influencing factor is cleaned sequentially. The cleaning process involves: using an anomaly detection algorithm to find outliers in each influencing factor, deleting the empirical samples corresponding to the outliers, and calculating the similarity using a similarity calculation formula.

[0019] The similarity calculation formula is as follows:

[0020]

[0021] in For similarity, Let be the difference between the number of iterations for the k-th empirical sample and the number of iterations for the target object. Let SoH be the difference between the k-th empirical sample and the target object.

[0022] Furthermore, the step of integrating the preset weight formula and the similarity to calculate the weight of each empirical sample includes:

[0023] The similarity is input into the weight formula to calculate the weight of each empirical sample;

[0024] The weighting formula is: ;in As weight, For similarity.

[0025] Furthermore, the step of iterating through the estimated remaining lifetime of each experience sample and calculating the prediction result for the target object using the weights and the estimated remaining lifetime includes:

[0026] The remaining lifetime of each empirical sample is read out, and the product of the remaining lifetime and the corresponding weight is added together to obtain the prediction result;

[0027] The prediction results are adjusted based on a preset frequency.

[0028] Furthermore, the method also includes:

[0029] Plot a fluctuation curve with time on the horizontal axis and the prediction result on the vertical axis, and insert labels generated by influencing factors into the fluctuation curve;

[0030] The system iterates through the abnormal features in the fluctuation curve, records the occurrence time of the abnormal features, finds the influencing factors at the occurrence time, obtains the target factors, integrates the abnormal features and target factors, generates feedback opinions, and sends them to the preset terminal.

[0031] Furthermore, the system includes:

[0032] The selection module is used to identify the target object for which new energy vehicle battery life prediction needs to be performed, determine the loading platform of the target object, read the attribute data of the target object, wherein the attribute data includes at least: battery code, cycle count and remaining service life, construct a usage database of batteries of the same brand, and select several experience samples from the usage database based on the cycle count, wherein the cycle count of the experience samples is greater than that of the target object;

[0033] The configuration module is used to configure the influencing factors of lifespan prediction, wherein the influencing factors include at least: temperature, charge and discharge rate and vehicle driving environment. The anomaly detection algorithm is used to clean the experience samples, construct a similarity calculation formula, and calculate the similarity between the target object and each experience sample in turn.

[0034] The sending module is used to integrate the preset weight formula and the similarity to calculate the weight of each experience sample, traverse the expected remaining life of each experience sample from the usage database, calculate the prediction result of the target object through the weight and the expected remaining life, and send the prediction result to the preset terminal.

[0035] Furthermore, the selection module includes:

[0036] The training unit is used to create a degradation model, collect historical operating data of the loading platform, and train the degradation model.

[0037] The correction unit is used to construct a time window, collect real-time operating condition data within the time window, input the real-time operating condition data and prediction results into the degradation model, and output a correction value to correct the prediction results.

[0038] A segmentation unit is used to segment each attribute data into several levels, where the target object and the experience sample correspond to the same level.

[0039] An adjustment unit is used to count the number of experience samples. When the number is less than a threshold, the level is adjusted, wherein the adjustment includes at least merging and re-segmenting.

[0040] Furthermore, the configuration module includes:

[0041] The cleaning unit is used to clean each influencing factor in turn. The cleaning process involves using an anomaly detection algorithm to find outliers in each influencing factor and delete the empirical samples corresponding to the outliers.

[0042] The calculation unit is used to calculate the similarity using a similarity calculation formula; the similarity calculation formula is: ;in For similarity, Let be the difference between the number of iterations for the k-th empirical sample and the number of iterations for the target object. Let SoH be the difference between the k-th empirical sample and the target object.

[0043] Compared with the prior art, the beneficial effects of the present invention are:

[0044] By identifying attribute data, personalized data can be provided for battery life prediction, improving prediction accuracy. By creating a usage database, the performance evolution process of similar batteries can be determined, enabling full life cycle management and providing a data foundation for battery life prediction. By identifying experience samples, a baseline for the target object can be established, avoiding prediction bias caused by battery differences and significantly improving the accuracy and stability of early battery life prediction. By calculating similarity and weight, the remaining usage time of the target object can be estimated using experience samples, effectively solving the problem of insufficient early data for new energy vehicle batteries and greatly improving prediction accuracy. Attached Figure Description

[0045] Figure 1 A flowchart illustrating the method for predicting the lifespan of new energy vehicle batteries provided in an embodiment of the present invention;

[0046] Figure 2 This is a first sub-flowchart of the new energy vehicle battery life prediction method provided in an embodiment of the present invention;

[0047] Figure 3 This is a second sub-flowchart of the new energy vehicle battery life prediction method provided in an embodiment of the present invention;

[0048] Figure 4 The third sub-flowchart of the new energy vehicle battery life prediction method provided in the embodiments of the present invention;

[0049] Figure 5 This is a block diagram of the new energy vehicle battery life prediction system provided in an embodiment of the present invention;

[0050] Figure 6 A block diagram showing the composition of selected modules in the new energy vehicle battery life prediction system provided in this embodiment of the invention;

[0051] Figure 7 A block diagram showing the composition of the configuration module in the new energy vehicle battery life prediction system provided in an embodiment of the present invention;

[0052] Figure 8 This is a block diagram of the sending module in the new energy vehicle battery life prediction system provided in an embodiment of the present invention. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0054] In Example 1, Figure 1The implementation flow of the new energy vehicle battery life prediction method provided in this embodiment of the invention is illustrated below, and is described in detail below:

[0055] S100: Identify the target object for which new energy vehicle battery life prediction needs to be performed, determine the loading platform of the target object, read the attribute data of the target object, wherein the attribute data includes at least: battery code, number of cycles and remaining service life, construct a usage database of batteries of the same brand, and select several experience samples from the usage database based on the number of cycles, wherein the number of cycles of the experience samples is greater than that of the target object.

[0056] In the field of new energy vehicle batteries, the target object for lifespan prediction is selected. This target object should be a relatively new new energy vehicle battery, in its early stages of use. The loading platform on which the target object is attached is determined, and this platform can be a car, truck, or other new energy vehicle with loading capabilities. The attribute data of the target object is then retrieved. This attribute data includes key indicators such as battery code, cycle count, and remaining lifespan. Specifically, the battery code uniquely identifies the individual battery, the cycle count reflects the battery's usage frequency, and the remaining lifespan characterizes the battery's current state and future usability. By determining this attribute data, a data foundation can be provided for subsequent maintenance decisions, energy efficiency assessments, and platform scheduling. In this application, the prior similar batteries refer to empirical samples.

[0057] Construct a database consisting of usage data of batteries of the same brand. The database contains historical operating data and status information of multiple batteries of the same brand. Read the cycle count of the target object and filter out several experience samples with a cycle count greater than that of the target object from the database. The experience samples can provide a performance trend reference for the target object in subsequent use. The experience samples refer to batteries with a cycle count greater than that of the target object.

[0058] S200: Configure the influencing factors of lifespan prediction, wherein the influencing factors include at least: temperature, charge and discharge rate and vehicle driving environment, use an anomaly detection algorithm to clean the experience samples, construct a similarity calculation formula, and calculate the similarity between the target object and each experience sample in turn.

[0059] Based on the actual usage of new energy batteries, factors affecting their lifespan are identified, including temperature and charge / discharge rate. Anomaly detection algorithms are used to remove outliers from each factor. For example, the operating temperature of each new energy vehicle battery is determined and sorted from highest to lowest temperature. The first quartile (Q1) and third quartile (Q3) are identified, and the interquartile range (IQR) is calculated. Batteries with operating temperatures less than Q1−1.5×IQR or greater than Q3+1.5×IQR are identified and their corresponding empirical samples are deleted. The process continues, sorting by charge / discharge rate from highest to lowest and deleting outliers from the empirical samples again. This process of deleting outliers is essentially cleaning the empirical samples. The similarity between the target battery and each cleaned empirical sample is calculated, with each empirical sample corresponding to a similarity score. A higher similarity score indicates that the battery's usage process is closer to (but greater than) that of the target battery, thus contributing more to the prediction of the target battery's lifespan.

[0060] S300: Integrate the preset weight formula and the similarity to calculate the weight of each experience sample, traverse the expected remaining lifespan of each experience sample from the usage database, calculate the prediction result of the target object through the weight and the expected remaining lifespan, and send the prediction result to the preset terminal.

[0061] The weight of each experience sample is calculated. The larger the weight, the greater the impact on the prediction of the remaining life of the target battery. The expected remaining life of each experience sample is read from the usage data. If the battery corresponding to the experience sample is also in the early stage of use, the corresponding experience sample is deleted. The experience sample and its corresponding weight are weighted and averaged. The average value is defined as the prediction result. The prediction result is used as the life prediction value of the target object. The prediction result is sent to a preset terminal, which can be a battery management personnel terminal.

[0062] In Example 2, Figure 2 The implementation flow of the new energy vehicle battery life prediction method provided by the embodiment of the present invention is shown. The following details the steps of identifying the target object for which new energy vehicle battery life prediction needs to be performed, determining the loading platform of the target object, and reading the attribute data of the target object:

[0063] S101: Create a degradation model, collect historical operating data of the loading platform, and train the degradation model.

[0064] Managers select appropriate data-driven models and use machine learning algorithms (such as regression analysis, support vector machines, random forests, etc.) to learn the degradation model, which is mainly used to predict the degradation process of new energy batteries. The degradation model is trained using historical operating data of the loading platform or new energy batteries.

[0065] S102: Construct a time window, collect real-time operating condition data within the time window, input the real-time operating condition data and prediction results into the degradation model, output the correction value, and correct the prediction results.

[0066] Set a time window (e.g., 1 hour, 1 day, or longer). Within each time window, monitor and collect operating condition data in real time. This data includes parameters related to battery degradation, such as temperature, charge / discharge rate, operating cycle, and mileage. Input the real-time operating condition data and prediction results into the degradation model. The degradation model will output correction values ​​based on the current operating condition data and historical information, combined with the degradation patterns learned during its training process. These correction values ​​can reduce the deviation between the prediction results and the actual degradation results.

[0067] In Example 3, Figure 2 The implementation flow of the new energy vehicle battery life prediction method provided by the embodiment of the present invention is shown. The following details the step of selecting several experience samples from the database:

[0068] S103: Divide each attribute data into several levels, where the target object and the experience sample correspond to the same level.

[0069] In each attribute data, select experience samples from the same level. For example, if there are four experience samples A, B, C, and D, taking the ambient temperature as an example, define below -15℃ as the first level, -15℃ to 0℃ as the second level, and 0℃ to 10℃ as the third level. Assuming that the target object, A, and C all correspond to the third level, while B and D correspond to the first and second levels respectively, then B and D will no longer be considered when selecting experience samples.

[0070] S104: Calculate the number of experience samples. When the number is less than a threshold, adjust the level. The adjustment includes at least merging and re-segmenting.

[0071] If the number of selected empirical samples is less than the threshold, the source of empirical samples can be increased by merging levels. In actual prediction, if the number of empirical samples is too small, it is easy to cause bias and reduce the accuracy of prediction.

[0072] In Example 4, Figure 3The implementation flow of the new energy vehicle battery life prediction method provided by an embodiment of the present invention is illustrated. The following details the steps of cleaning the experience samples using an anomaly detection algorithm, constructing a similarity calculation formula, and sequentially calculating the similarity between the target object and each experience sample, as follows:

[0073] S201: Clean each influencing factor in turn, wherein the cleaning is: using an anomaly detection algorithm, finding outliers in each influencing factor and deleting the empirical samples corresponding to the outliers.

[0074] For each influencing factor, the portion corresponding to values ​​less than Q1−1.5×IQR or greater than Q3+1.5×IQR is defined as an outlier, and the empirical samples corresponding to the outliers are deleted.

[0075] S202: Calculate the similarity using the similarity calculation formula; the similarity calculation formula is as follows: ;in For similarity, Let be the difference between the number of iterations for the k-th empirical sample and the number of iterations for the target object. Let SoH be the difference between the k-th empirical sample and the target object.

[0076] The data such as the number of cycles between each experience sample and the target object, SoH, and remaining useful life are read from the attribute data; and then substituted into the similarity calculation formula to calculate the similarity between each experience sample and the target object; the denominator of the similarity calculation formula may also include other attribute data, which are selected by the management personnel.

[0077] In Example 5, Figure 4 The implementation flow of the new energy vehicle battery life prediction method provided by the embodiment of the present invention is illustrated. The following details the steps of integrating the preset weight formula and the similarity to calculate the weight of each empirical sample:

[0078] S301: Input the similarity into the weighting formula to calculate the weight of each empirical sample. The weighting formula is as follows: ;in As weight, For similarity.

[0079] The similarity of each empirical sample is input into the weight formula to calculate the weight.

[0080] In Example 6, Figure 4 The implementation flow of the new energy vehicle battery life prediction method provided by an embodiment of the present invention is illustrated. The following details the steps of iterating through the expected remaining life of each experience sample and calculating the prediction result of the target object using the weight and the expected remaining life:

[0081] S302: Read the remaining lifetime of each empirical sample and sum the product of the remaining lifetime and the corresponding weight to obtain the prediction result.

[0082] The remaining lifetime of each experience sample is multiplied by its corresponding weight to obtain the product corresponding to that experience sample. All products are then summed to obtain the prediction result.

[0083] S303: Adjust the prediction results based on a preset frequency.

[0084] The prediction results are updated based on a preset frequency, which can be set by the administrator.

[0085] In Example 7, unlike Example 1, the method further includes:

[0086] Plot a fluctuation curve with time on the horizontal axis and the prediction result on the vertical axis, and insert labels generated by influencing factors into the fluctuation curve;

[0087] The system iterates through the abnormal features in the fluctuation curve, records the occurrence time of the abnormal features, finds the influencing factors at the occurrence time, obtains the target factors, integrates the abnormal features and target factors, generates feedback opinions, and sends them to the preset terminal.

[0088] Plot a fluctuation curve with time on the x-axis and prediction result on the y-axis. The fluctuation curve reflects the changing trend of the prediction result. Traverse the fluctuation curve to identify abnormal feature points, such as sudden drops, sudden rises, or fluctuations that do not conform to the normal degradation trend. Record the occurrence time corresponding to each abnormal feature, and use this time point as an index to retrieve and extract the real-time operating condition data collected at that time point. Further analyze the influencing factors that lead to abnormal features, such as abnormal charging and discharging rates.

[0089] Based on the influencing factors, the specific causes of the anomalies are determined, such as high-rate charging and discharging, battery damage, and battery management system malfunctions. Based on the causes of the anomalies, targeted feedback is generated, such as adjusting the charging strategy, checking the cooling system, or optimizing driving behavior. The feedback is sent to a preset terminal in the form of text or charts, where the preset terminal is the management personnel terminal.

[0090] Figure 5 This diagram illustrates the structural block diagram of a new energy vehicle battery life prediction system provided in an embodiment of the present invention. The new energy vehicle battery life prediction system 1 includes:

[0091] The selection module 11 is used to identify the target object for which new energy vehicle battery life prediction needs to be performed, determine the loading platform of the target object, read the attribute data of the target object, wherein the attribute data includes at least: battery code, cycle number and remaining service life, construct a usage database of batteries of the same brand, and select several experience samples from the usage database based on the cycle number, wherein the cycle number of the experience samples is greater than that of the target object.

[0092] Configuration module 12 is used to configure the influencing factors of lifespan prediction, wherein the influencing factors include at least: temperature, charge and discharge rate and vehicle driving environment. Using an anomaly detection algorithm, the experience samples are cleaned, a similarity calculation formula is constructed, and the similarity between the target object and each experience sample is calculated in turn.

[0093] The sending module 13 is used to integrate the preset weight formula and the similarity to calculate the weight of each experience sample, traverse the expected remaining life of each experience sample from the usage database, calculate the prediction result of the target object through the weight and the expected remaining life, and send the prediction result to the preset terminal.

[0094] Figure 6 This diagram illustrates the structural block diagram of a new energy vehicle battery life prediction system provided in an embodiment of the present invention. The selection module 11 includes:

[0095] Training unit 111 is used to create a degradation model, collect historical operating data of the loading platform, and train the degradation model.

[0096] The correction unit 112 is used to construct a time window, collect real-time operating condition data within the time window, input the real-time operating condition data and prediction results into the degradation model, and output a correction value to correct the prediction results.

[0097] The segmentation unit 113 is used to segment each attribute data into several levels, wherein the target object and the experience sample correspond to the same level.

[0098] The adjustment unit 114 is used to count the number of experience samples. When the number is less than a threshold, the level is adjusted, wherein the adjustment includes at least merging and re-segmenting.

[0099] Figure 7 This diagram illustrates the structural block diagram of a new energy vehicle battery life prediction system provided in an embodiment of the present invention. The configuration module 12 includes:

[0100] The cleaning unit 121 is used to clean each influencing factor in sequence, wherein the cleaning is: using an anomaly detection algorithm, finding outliers in each influencing factor and deleting the empirical samples corresponding to the outliers;

[0101] Calculation unit 122 is used to calculate the similarity using a similarity calculation formula; the similarity calculation formula is: ;in For similarity, Let be the difference between the number of iterations for the k-th empirical sample and the number of iterations for the target object. Let SoH be the difference between the k-th empirical sample and the target object.

[0102] Figure 8 This diagram illustrates the structural block diagram of a new energy vehicle battery life prediction system provided in an embodiment of the present invention. The transmitting module 13 includes:

[0103] Input unit 131 is used to input the similarity into the weight formula to calculate the weight of each empirical sample. The weight formula is as follows: ;in As weight, Similarity;

[0104] Unit 132 is used to read the remaining lifespan of each experience sample and superimpose the product of the remaining lifespan and the corresponding weight to obtain the prediction result;

[0105] The adjustment unit 133 is used to adjust the prediction result according to a preset frequency.

[0106] The selection module 11 is mainly used to complete step S100, the configuration module 12 is mainly used to complete step S200, and the sending module 13 is mainly used to complete step S300.

[0107] The training unit 111 is mainly used to complete step S101, the correction unit 112 is mainly used to complete step S102, the segmentation unit 113 is mainly used to complete step S103, and the adjustment unit 114 is mainly used to complete step S104.

[0108] The cleaning unit 121 is mainly used to complete step S201, and the calculation unit 122 is mainly used to complete step S202.

[0109] The input unit 131 is mainly used to complete step S301, the obtaining unit 132 is mainly used to complete step S302, and the adjustment unit 133 is mainly used to complete step S303.

[0110] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0111] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0112] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting the lifespan of new energy vehicle batteries, characterized in that, The method includes: The target object for predicting the battery life of new energy vehicles is identified, the loading platform of the target object is determined, and the attribute data of the target object is read. The attribute data includes at least the battery code, cycle count and remaining service life. A usage database of batteries of the same brand is constructed. Based on the cycle count, several experience samples are selected from the usage database, wherein the cycle count of the experience samples is greater than that of the target object. The factors influencing lifespan prediction are configured, including at least temperature, charge / discharge rate and vehicle driving environment. An anomaly detection algorithm is used to clean the experience samples, a similarity calculation formula is constructed, and the similarity between the target object and each experience sample is calculated sequentially. By integrating the preset weight formula and the similarity, the weight of each experience sample is calculated. From the usage database, the expected remaining lifespan of each experience sample is traversed. Based on the weight and the expected remaining lifespan, the prediction result of the target object is calculated and sent to the preset terminal. The steps of cleaning the experience samples using an anomaly detection algorithm, constructing a similarity calculation formula, and sequentially calculating the similarity between the target object and each experience sample include: Each influencing factor is cleaned sequentially. The cleaning process involves: using an anomaly detection algorithm to find outliers in each influencing factor, deleting the empirical samples corresponding to the outliers, and calculating the similarity using a similarity calculation formula. The similarity calculation formula is as follows: ； in Let be the similarity between the k-th empirical sample and the target object. Let be the difference between the number of iterations for the k-th empirical sample and the number of iterations for the target object. Let SoH be the difference between the k-th empirical sample and the target object.

2. The method for predicting the lifespan of new energy vehicle batteries according to claim 1, characterized in that, The steps of identifying the target object for which new energy vehicle battery life prediction needs to be performed, determining the loading platform of the target object, and reading the attribute data of the target object include: A degradation model is created by collecting historical operating data of the loading platform and training the degradation model. A time window is constructed, real-time operating condition data within the time window is collected, the real-time operating condition data and the prediction results are input into the degradation model, and the correction value is output to correct the prediction results.

3. The method for predicting the lifespan of new energy vehicle batteries according to claim 1, characterized in that, The step of selecting several experience samples from the usage database includes: Each attribute data is divided into several levels, where the target object and the experience sample correspond to the same level; The number of experience samples is counted. When the number is less than a threshold, the level is adjusted, and the adjustment includes at least merging and re-segmenting.

4. The method for predicting the lifespan of new energy vehicle batteries according to claim 1, characterized in that, The step of integrating the preset weight formula and the similarity to calculate the weight of each empirical sample includes: The similarity is input into the weight formula to calculate the weight of each empirical sample; The weighting formula is: ; in As weight, For similarity.

5. The method for predicting the lifespan of new energy vehicle batteries according to claim 4, characterized in that, The step of iterating through the estimated remaining lifetime of each experience sample and calculating the prediction result of the target object using the weights and the estimated remaining lifetime includes: The remaining lifetime of each empirical sample is read out, and the product of the remaining lifetime and the corresponding weight is added together to obtain the prediction result; The prediction results are adjusted based on a preset frequency.

6. The method for predicting the lifespan of new energy vehicle batteries according to claim 1, characterized in that, The method further includes: Plot a fluctuation curve with time on the horizontal axis and the prediction result on the vertical axis, and insert labels generated by influencing factors into the fluctuation curve; The system iterates through the abnormal features in the fluctuation curve, records the occurrence time of the abnormal features, finds the influencing factors at the occurrence time, obtains the target factors, integrates the abnormal features and target factors, generates feedback opinions, and sends them to the preset terminal.

7. A new energy vehicle battery life prediction system, characterized in that, The system includes: The selection module is used to identify the target object for which new energy vehicle battery life prediction needs to be performed, determine the loading platform of the target object, read the attribute data of the target object, wherein the attribute data includes at least: battery code, cycle count and remaining service life, construct a usage database of batteries of the same brand, and select several experience samples from the usage database based on the cycle count, wherein the cycle count of the experience samples is greater than that of the target object; The configuration module is used to configure the influencing factors of lifespan prediction, wherein the influencing factors include at least: temperature, charge and discharge rate and vehicle driving environment. The anomaly detection algorithm is used to clean the experience samples, construct a similarity calculation formula, and calculate the similarity between the target object and each experience sample in turn. The sending module is used to integrate the preset weight formula and the similarity to calculate the weight of each experience sample, traverse the expected remaining life of each experience sample from the usage database, calculate the prediction result of the target object through the weight and the expected remaining life, and send the prediction result to the preset terminal. The configuration module includes: The cleaning unit is used to clean each influencing factor in turn. The cleaning process involves using an anomaly detection algorithm to find outliers in each influencing factor and delete the empirical samples corresponding to the outliers. The calculation unit is used to calculate the similarity using a similarity calculation formula; the similarity calculation formula is: ;in For similarity, Let be the difference between the number of iterations for the k-th empirical sample and the number of iterations for the target object. Let SoH be the difference between the k-th empirical sample and the target object.

8. The new energy vehicle battery life prediction system according to claim 7, characterized in that, The selection module includes: The training unit is used to create a degradation model, collect historical operating data of the loading platform, and train the degradation model. The correction unit is used to construct a time window, collect real-time operating condition data within the time window, input the real-time operating condition data and prediction results into the degradation model, and output a correction value to correct the prediction results. A segmentation unit is used to segment each attribute data into several levels, where the target object and the experience sample correspond to the same level. An adjustment unit is used to count the number of experience samples. When the number is less than a threshold, the level is adjusted, wherein the adjustment includes at least merging and re-segmenting.

Images