Mutual Learning Prediction Method, Device and Storage Medium for Vehicle Battery Life

Abstract

The present invention discloses a mutual learning prediction method, device and storage medium for the battery life of a vehicle. The method includes: determining an empirical model, and obtaining the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery; performing mutual learning on the historical data trajectories of the vehicle's own and other vehicle batteries by using the empirical model; estimating the battery capacity state based on big data to obtain a first estimation result, and correcting the parameters of the empirical model according to the first estimation result; determining a mechanism model, and estimating the battery capacity state according to the mechanism model and big data to obtain a second estimation result; predicting the battery life by using the corrected empirical model, and correcting the parameters of the corrected empirical model again according to the second estimation result, and finally obtaining the battery life prediction result. Thus, a more accurate battery life prediction model can be constructed, and the battery life prediction accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of batteries, and in particular, to a mutual learning prediction method, device, and storage medium for the life of vehicle batteries. Background Art

[0002] With the rapid development of electric vehicles, higher requirements are put forward for power batteries. The service life of power batteries will directly affect the performance of electric vehicles. Therefore, people pay more attention to the research on the prediction of the service life of power batteries. Accurate battery life prediction can not only improve the user driving experience, but also build a dynamic intelligent health management system for the whole life cycle of power batteries, with great social and economic benefits.

[0003] In the related art, when predicting the service life of power batteries, the prediction of the service life of power batteries is usually based on data-driven historical trajectories, or the prediction of the service life of power batteries is based on a mechanism model. However, these common methods for predicting the service life of power batteries are difficult to establish a relatively accurate prediction model, and the accuracy of battery life prediction is poor. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems in the related art to some extent. To this end, the first object of the present invention is to propose a mutual learning prediction method for the life of vehicle batteries, which successively corrects the determined empirical model parameters twice based on big data and a determined mechanism model, so as to build a more accurate battery life prediction model and improve the accuracy of battery life prediction.

[0005] The second object of the present invention is to propose a computer-readable storage medium.

[0006] The third object of the present invention is to propose a cloud server.

[0007] The fourth object of the present invention is to propose a mutual learning prediction device for the life of vehicle batteries.

[0008] To achieve the above object, an embodiment of the first aspect of the present invention provides a mutual learning prediction method for vehicle battery life, including: determining an empirical model, obtaining the historical data trajectory of the vehicle's own battery, and obtaining the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery; using the empirical model to perform mutual learning on the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries; estimating the battery capacity state based on big data to obtain a first estimation result, and during the mutual learning process, correcting the parameters of the empirical model according to the first estimation result; determining a mechanism model, and estimating the battery capacity state according to the mechanism model and big data to obtain a second estimation result; using the corrected empirical model to predict the battery life, and during the prediction process, correcting the parameters of the corrected empirical model again according to the second estimation result, and finally obtaining the battery life prediction result.

[0009] According to the mutual learning prediction method for vehicle battery life of the embodiment of the present invention, an empirical model is determined, the historical data trajectory of the vehicle's own battery is obtained, and the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery is obtained. The empirical model is used to perform mutual learning on the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries. The battery capacity state is estimated based on big data to obtain a first estimation result, and during the mutual learning process, the parameters of the empirical model are corrected according to the first estimation result. A mechanism model is determined, and the battery capacity state is estimated according to the mechanism model and big data to obtain a second estimation result. The corrected empirical model is used to predict the battery life, and during the prediction process, the parameters of the corrected empirical model are corrected again according to the second estimation result, and finally the battery life prediction result is obtained. Thus, based on big data and the determined mechanism model, the parameters of the determined empirical model are corrected twice successively, so that a more accurate battery life prediction model can be constructed, and the battery life prediction accuracy is improved.

[0010] According to an embodiment of the present invention, obtaining the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery includes: when the battery type of other vehicles is the same as that of the vehicle's own battery, if the battery cycle times of other vehicles are greater than those of the vehicle's own battery, the battery life prediction values of other vehicles are less than those of the vehicle's own battery, and the battery attenuation rates of other vehicles and the vehicle's own battery meet a preset condition, it is determined that the historical data trajectory of other vehicle batteries is adapted to the historical data trajectory of the vehicle's own battery.

[0011] According to an embodiment of the present invention, when the following relational expression is satisfied between the battery attenuation rate of other vehicles and the battery attenuation rate of the vehicle's own battery, it is determined that the battery attenuation rate of other vehicles and the battery attenuation rate of the vehicle's own battery meet the preset condition: Wherein, VSOHCi is the battery attenuation rate of other vehicles, V SOHC is the attenuation rate of the vehicle's own battery.

[0012] According to an embodiment of the present invention, modifying the parameters of the empirical model according to the first estimation result includes: using the first learning result as a prior estimate, and using the first estimation result as a posterior correction, and using the Kalman filter algorithm to modify the parameters of the empirical model, where the first learning result is obtained by learning the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicles' batteries according to the empirical model.

[0013] According to an embodiment of the present invention, modifying the parameters of the modified empirical model again according to the second estimation result includes: using the second learning result output by the modified empirical model as a prior estimate, and using the second estimation result as a posterior correction, and using the Kalman filter algorithm to modify the parameters of the modified empirical model again.

[0014] According to an embodiment of the present invention, estimating the battery capacity state based on big data to obtain a first estimation result includes: obtaining a first life prediction value of the vehicle's own battery and a life prediction value of other vehicles' batteries at the same number of cycles; performing weighted calculation according to the first life prediction value of the vehicle's own battery and the life prediction value of other vehicles' batteries to obtain the first estimation result.

[0015] According to an embodiment of the present invention, the first estimation result is calculated according to the following formula:

[0016]

[0017] where, when V SOHCi ≥V SOHC when, when V SOHCi <V SOHC when, SOHC1 is the first estimation result, N is the number of vehicles that meet the screening conditions, P i is the weight coefficient, SOHC i is the life prediction value of other vehicles' batteries, and SOHCself1 is the first life prediction value of the vehicle's own battery.

[0018] According to an embodiment of the present invention, estimating the battery capacity state according to the mechanism model and big data to obtain a second estimation result includes: obtaining a second life prediction value of the vehicle's own battery at the first preset number of cycles; obtaining a life prediction change value of the vehicle's own battery at the first preset number of cycles according to the mechanism model and big data; performing addition calculation according to the second life prediction value of the vehicle's own battery and the life prediction change value of the vehicle's own battery to obtain the second estimation result.

[0019] According to an embodiment of the present invention, the second estimation result is calculated according to the following formula:

[0020] SOHC2 = SOHCself2 + ΔSOHC,

[0021] where SOHC2 is the second estimation result, SOHCself2 is the predicted second life of the vehicle's own battery, and ΔSOHC is the predicted change in the life of the vehicle's own battery.

[0022] According to an embodiment of the present invention, the predicted change in the life of the vehicle's own battery at the first preset number of cycles is obtained based on the mechanism model and big data, including: obtaining the predicted change in the life of the battery at the second preset number of cycles of other vehicles according to big data, where the second preset number of cycles is the number of cycles by which the battery of each vehicle among other vehicles has decayed within the first preset number of cycles; obtaining the predicted change in the life of the battery at the second preset number of cycles of the vehicle itself according to the mechanism model, and the predicted change in the life of the battery at the third preset number of cycles of the vehicle itself, where the third preset number of cycles is the number of cycles by which the battery of each vehicle among other vehicles has not decayed within the first preset number of cycles; first performing weighted calculation according to the predicted change in the life of the battery at the second preset number of cycles of other vehicles and the predicted change in the life of the battery at the second preset number of cycles of the vehicle itself, and adding the weighted calculation result to the predicted change in the life of the battery at the third preset number of cycles of the vehicle itself to obtain the predicted change in the life of the vehicle's own battery.

[0023] According to an embodiment of the present invention, the predicted change in the life of the vehicle's own battery is calculated according to the following formula:

[0024]

[0025] where, when V SOHCi ≥V SOHC then when V SOHCi <V SOHC then ΔSOHC is the predicted change in the life of the vehicle's own battery, N is the number of vehicles meeting the screening conditions, P i is the weight coefficient, ΔSOHCD i is the predicted change in the life of the battery at the second preset number of cycles of other vehicles, ΔSOHCM is the predicted change in the life of the battery at the second preset number of cycles of the vehicle itself, and ΔSOHCMelse is the predicted change in the life of the battery at the third preset number of cycles of the vehicle itself.

[0026] To achieve the above object, an embodiment of the second aspect of the present invention provides a computer-readable storage medium, on which a mutual learning prediction program for vehicle battery life is stored. When the mutual learning prediction program for vehicle battery life is executed by a processor, the mutual learning prediction method for vehicle battery life in the embodiment of the first aspect is implemented.

[0027] According to the computer-readable storage medium of the embodiment of the present invention, through the above mutual learning prediction method for vehicle battery life, based on big data and a determined mechanism model, the determined empirical model parameters are corrected twice successively, so that a more accurate battery life prediction model can be constructed, and the battery life prediction accuracy is improved.

[0028] To achieve the above object, an embodiment of the third aspect of the present invention provides a cloud server, including a memory, a processor, and a mutual learning prediction program for vehicle battery life stored on the memory and executable on the processor. When the processor executes the mutual learning prediction program for vehicle battery life, the mutual learning prediction method for vehicle battery life in the embodiment of the first aspect is implemented.

[0029] According to the cloud server of the embodiment of the present invention, through the above mutual learning prediction method for vehicle battery life, based on big data and a determined mechanism model, the determined empirical model parameters are corrected twice successively, so that a more accurate battery life prediction model can be constructed, and the battery life prediction accuracy is improved.

[0030] To achieve the above object, an embodiment of the fourth aspect of the present invention provides a mutual learning prediction device for vehicle battery life, including: a first determination module, configured to determine an empirical model, obtain the historical data trajectory of the vehicle's own battery, and obtain the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery, and perform mutual learning on the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries by using the empirical model; a first estimation module, configured to estimate the battery capacity state based on big data to obtain a first estimation result; a first correction module, configured to correct the parameters of the empirical model according to the first estimation result during the mutual learning process; a second determination module, configured to determine a mechanism model; a second estimation module, configured to estimate the battery capacity state according to the mechanism model and big data to obtain a second estimation result; and a prediction module, configured to predict the battery life by using the corrected empirical model, and during the prediction process, correct the parameters of the corrected empirical model again according to the second estimation result to finally obtain a battery life prediction result.

[0031] The mutual learning prediction device for the vehicle battery life according to an embodiment of the present invention determines an empirical model through a first determination module, obtains the historical data trajectory of the vehicle's own battery, and obtains the historical data trajectories of other vehicle batteries adapted to the vehicle's own battery, and performs mutual learning on the historical data trajectory of the vehicle's own battery and the historical data trajectories of other vehicle batteries by using the empirical model. The first estimation module estimates the battery capacity state based on big data to obtain a first estimation result. During the mutual learning process, the first correction module corrects the parameters of the empirical model according to the first estimation result. The second determination module determines a mechanism model, and the second estimation module estimates the battery capacity state according to the mechanism model and big data to obtain a second estimation result. The prediction module predicts the battery life by using the corrected empirical model, and during the prediction process, corrects the parameters of the corrected empirical model again according to the second estimation result, and finally obtains the battery life prediction result. Thus, based on big data and the determined mechanism model, the parameters of the determined empirical model are corrected twice successively, so as to construct a more accurate battery life prediction model and improve the battery life prediction accuracy.

[0032] Additional aspects and advantages of the present invention will be given in part in the following description, will become apparent in part from the following description, or will be understood through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a flowchart of a mutual learning prediction method for the vehicle battery life according to an embodiment of the present invention;

[0034] Figure 2 It is a schematic flow diagram of obtaining the vehicle battery life prediction result according to an embodiment of the present invention;

[0035] Figure 3 It is a flowchart of obtaining the second estimation result according to an embodiment of the present invention;

[0036] Figure 4 It is a schematic structural diagram of a mutual learning prediction device for the vehicle battery life according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the drawings, in which the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions from beginning to end. The embodiments described below with reference to the drawings are exemplary and are intended to explain the present invention, but should not be construed as limiting the present invention.

[0038] The mutual learning prediction method, device, cloud server and computer-readable storage medium for the vehicle battery life proposed by the embodiments of the present invention will be described below with reference to the drawings.

[0039] Figure 1 It is a flowchart of a mutual learning prediction method for the life of a vehicle battery according to an embodiment of the present invention. As Figure 1 shown, the mutual learning prediction method for the life of the vehicle battery includes the following steps:

[0040] Step S101, determine an empirical model, obtain the historical data trajectory of the vehicle's own battery, and obtain the historical data trajectories of other vehicle batteries adapted to the vehicle's own battery.

[0041] It should be noted that since different batteries have different positive and negative components, different proportions, and different manufacturing processes during the manufacturing process, and the batteries will experience different usage conditions during use, the capacity decay trajectories finally shown by the batteries are significantly different. Therefore, when predicting the battery life through an empirical model, it is necessary to determine a suitable empirical model to make the prediction result of the battery life as accurate as possible.

[0042] Specifically, when determining the empirical model for predicting the battery life, an offline experiment is carried out on the battery to be tested to determine the type, usage conditions, and corresponding aging mode of the battery. Based on the type, usage conditions, and corresponding aging mode of the above battery, the mean square error under different empirical models is obtained, and the minimum mean square error among different empirical models is determined. Based on the obtained minimum mean square error, the empirical model corresponding to the minimum mean square error is selected as the empirical model to be adopted. Currently, common empirical models include the double exponential model, single exponential model, linear model, polynomial model, and Verhulst model. That is to say, a suitable empirical model can be selected from the double exponential model, single exponential model, linear model, polynomial model, and Verhulst model to ensure that the mean square error value corresponding to the selected empirical model is the smallest, and the empirical model corresponding to the minimum mean square error is used as the empirical model to be adopted. For example, when the mean square error value calculated by the double exponential model among different empirical models is the smallest, the double exponential model is used as the empirical model to be adopted.

[0043] Currently, with the rapid development of technologies such as big data, cloud platforms, and intelligent algorithms, using a database to store and manage a large amount of historical data can realize the mutual learning and optimization of different vehicle parameters, thereby further improving the accuracy of battery life prediction. Therefore, by obtaining the historical data trajectory of the vehicle's own battery and the historical data trajectories of other vehicle batteries adapted to the vehicle's own battery, the number of historical data trajectories in the database can be expanded.

[0044] In some embodiments, obtaining the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery includes: when the battery type of other vehicles is the same as that of the vehicle's own battery, if the number of battery cycles of other vehicles is greater than that of the vehicle's own battery, the predicted battery life value of other vehicles is less than that of the vehicle's own battery, and the battery attenuation rate of other vehicles and the battery attenuation rate of the vehicle's own battery meet the preset conditions, then it is determined that the historical data trajectory of other vehicle batteries is adapted to the historical data trajectory of the vehicle's own battery.

[0045] Further, when the following relational expression is satisfied between the battery attenuation rate of other vehicles and the battery attenuation rate of the vehicle's own battery, it is determined that the battery attenuation rate of other vehicles and the battery attenuation rate of the vehicle's own battery meet the preset conditions: where V SOHCi is the battery attenuation rate of other vehicles, and V SOHC is the battery attenuation rate of the vehicle's own battery.

[0046] Specifically, in order to increase the number of historical data trajectories in the database, when obtaining the historical data trajectory of the vehicle's own battery, it is also necessary to obtain the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery. When obtaining the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery, first determine whether the battery type of other vehicles is the same as that of the vehicle's own battery. If the battery type of other vehicles is different from that of the vehicle's own battery, then do not obtain the historical data trajectory of the vehicle corresponding to this battery. If the battery type of other vehicles is the same as that of the vehicle's own battery, then determine whether the number of battery cycles of other vehicles is greater than that of the vehicle's own battery, whether the predicted battery life value of other vehicles is less than that of the vehicle's own battery, and whether the battery attenuation rate of other vehicles and the battery attenuation rate of the vehicle's own battery meet the preset conditions: where V SOHCi is the battery attenuation rate of other vehicles, and V SOHC is the battery attenuation rate of the vehicle's own battery. If other vehicles meet the above three conditions at the same time, it means that the historical data trajectory of other vehicle batteries is adapted to the historical data trajectory of the vehicle's own battery, and the historical data trajectory of other vehicle batteries and the historical data trajectory of the vehicle's own battery can be stored together in the historical data trajectory database.

[0047] Step S102, perform mutual learning on the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries using an empirical model.

[0048] Specifically, after obtaining the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery, the determined empirical model is used to perform mutual learning on the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries. Since the historical data trajectory used in the empirical model includes the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery, the database for mutual learning is greatly expanded, which is beneficial to more accurately correcting the parameters of the empirical model.

[0049] Step S103, estimating the battery capacity state based on big data to obtain a first estimation result, and during the mutual learning process, correcting the parameters of the empirical model according to the first estimation result.

[0050] Specifically, the battery capacity is an important performance index for evaluating the battery after long-term use. The battery capacity refers to the total amount of charge generated during the complete discharge process of the battery under given conditions and time. During the repeated charging process of the battery, the battery capacity will decay. Therefore, during the battery life prediction process, it is necessary to consider the impact of the current battery capacity state on the battery life assessment.

[0051] In some embodiments, estimating the battery capacity state based on big data to obtain a first estimation result includes: obtaining the first life prediction value of the vehicle's own battery and the life prediction value of other vehicle batteries at the same number of cycles; performing weighted calculation according to the first life prediction value of the vehicle's own battery and the life prediction value of other vehicle batteries to obtain the first estimation result.

[0052] It should be noted that estimating the battery capacity state based on big data means obtaining the life prediction values at the same number of cycles for the vehicle itself and other vehicles respectively in a historical big data-based manner. Among them, obtaining the life prediction value of the vehicle battery based on historical big data is as follows: respectively establishing the mapping relationship between the characteristic voltage segment at the initial stage of the vehicle and the battery capacity and the mapping relationship between the current characteristic voltage segment of the vehicle and the battery capacity according to the historical voltage data and battery capacity data of the vehicle battery to be predicted, that is, obtaining the battery capacity of the characteristic voltage segment at the initial stage of the vehicle and the battery capacity of the current characteristic voltage segment of the vehicle. The battery capacity of the characteristic voltage segment at the initial stage of the vehicle obtained according to the mapping relationship between the characteristic voltage segment at the initial stage of the vehicle and the battery capacity, and the battery capacity of the current characteristic voltage segment of the vehicle obtained according to the mapping relationship between the current characteristic voltage segment of the vehicle and the battery capacity are used to estimate the current vehicle battery life prediction value. The estimation formula for the vehicle battery life prediction value based on historical big data is shown as follows:

[0053]

[0054] Among them, SOHCL is the predicted value of the vehicle battery life estimated based on historical big data, and C now is the current battery capacity of the vehicle; C new is the battery capacity at the initial stage of the vehicle, and C featurenow is the battery capacity of the current characteristic voltage segment of the vehicle; C featurenew is the battery capacity of the characteristic voltage segment at the initial stage of the vehicle.

[0055] According to the above method for obtaining the predicted value of the vehicle battery life based on historical big data, the first life prediction value of the vehicle's own battery and the life prediction values of other vehicle batteries are respectively obtained at the same number of cycles, and weighted calculation is performed based on the first life prediction value of the vehicle's own battery and the life prediction values of other vehicle batteries to obtain the first estimation result.

[0056] Furthermore, the formula for obtaining the first estimation result by weighted calculation based on the first life prediction value of the vehicle's own battery and the life prediction values of other vehicle batteries is specifically as follows:

[0057]

[0058] Among them, when V SOHCi ≥V SOHC ; when V SOHCi <V SOHC ; SOHC1 is the first estimation result, N is the number of vehicles that meet the screening conditions, P i is the weight coefficient, SOHC i is the life prediction value of other vehicle batteries, and SOHCself1 is the first life prediction value of the vehicle's own battery.

[0059] It should be noted that the life prediction value SOHC i of other vehicle batteries and the first life prediction value SOHCself1 of the vehicle's own battery are both obtained based on the above historical big data, and the specific formulas are respectively expressed as follows:

[0060]

[0061] Among them, SOHC i is the life prediction value of other vehicle batteries, C nowi is the current battery capacity of other vehicles; C newi is the battery capacity at the initial stage of other vehicles, C featurenowi is the battery capacity of the current characteristic voltage segment of other vehicles; C featurenewi is the battery capacity of the characteristic voltage segment at the initial stage of other vehicles.

[0062]

[0063] Among them, SOHCself1 is the first life prediction value of the vehicle's own battery, and C nowself1 is the first current battery capacity of the vehicle itself; C newself is the initial-stage battery capacity of the vehicle itself, and C featurenowself1 is the battery capacity of the first current characteristic voltage segment of the vehicle itself; C featurenewself is the battery capacity of the initial-stage characteristic voltage segment of the vehicle itself.

[0064] In some embodiments, the parameters of the empirical model are corrected according to the first estimation result, including: taking the first learning result as a prior estimate and the first estimation result as a posterior correction, and using the Kalman filter algorithm to correct the parameters of the empirical model. Among them, the first learning result is obtained by learning the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicles' batteries according to the empirical model.

[0065] Specifically, as Figure 2 shown, first, the battery capacity state is estimated based on big data to obtain the first estimation result SOHC1. And in the battery life prediction process of the learning stage, the first learning result obtained by learning the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicles' batteries according to the empirical model is taken as a prior estimate, and the first estimation result SOHC1 is taken as a posterior correction. The Kalman filter algorithm is used to perform the first correction on the parameters of the determined empirical model according to the input first learning result and the first estimation result, and the empirical model with the first corrected parameters is used to predict the battery life to improve the battery life prediction accuracy.

[0066] Step S104, determine the mechanism model, and estimate the battery capacity state according to the mechanism model and big data to obtain a second estimation result.

[0067] It should be noted that this application also uses a mechanism model to predict the battery capacity state. Similar to determining a suitable empirical model, before using the mechanism model, a suitable mechanism model needs to be selected. Commonly used mechanism models include: single-particle model, P2D electrochemical model, and electro-thermal-mechanical coupling model. The suitable mechanism model is determined offline according to the type of battery, usage conditions, and corresponding aging mode.

[0068] Specifically, as Figure 3 shown, in some embodiments, estimating the battery capacity state according to the mechanism model and big data to obtain a second estimation result includes:

[0069] Step S201, obtain the second life prediction value of the vehicle's own battery at the first preset number of cycles.

[0070] Specifically, the second life prediction value of the vehicle's own battery when obtaining the first preset number of cycles is obtained by obtaining the second life prediction value of the vehicle's own battery when the first preset number of cycles is based on historical big data. The acquisition formula for the second life prediction value of the vehicle's own battery is specifically expressed as follows:

[0071]

[0072] Among them, SOHCself2 is the second life prediction value of the vehicle's own battery, C nowself2 is the second current battery capacity of the vehicle itself; C newself is the battery capacity at the initial stage of the vehicle itself, C featurenowself2 is the battery capacity of the second current characteristic voltage segment of the vehicle itself; C featurenewself is the battery capacity of the initial stage characteristic voltage segment of the vehicle itself.

[0073] Step S202: Obtain the predicted change value of the life of the vehicle's own battery when the first preset number of cycles is reached according to the mechanism model and big data.

[0074] It should be noted that to determine the predicted change value of the battery life of the vehicle, it is first necessary to obtain the predicted battery life values of the vehicle battery according to the mechanism model and big data respectively. Among them, the estimation of the predicted battery life value of the vehicle battery based on the mechanism model includes: based on the Butler-Volmer equation, establish the relationship between the side reaction rate of the battery electrode and the overpotential of the side reaction, and determine the battery capacity loss according to the relationship between the side reaction rate of the battery electrode and the overpotential of the side reaction. The battery capacity loss can be specifically expressed as:

[0075]

[0076] Among them, Q loss，cyc is the battery capacity loss of the vehicle battery, F is the Faraday constant, t cyc is the time for 1 cycle, is the loss of lithium ions, and T is the number of cycles.

[0077] Estimate the battery capacity state according to the battery capacity loss, that is, obtain the predicted battery life value of the vehicle. The battery capacity state can be specifically expressed as:

[0078]

[0079] Among them, SOHCJ is the predicted battery life value of the vehicle estimated based on the mechanism model, Q loss，cyc is the battery capacity loss of the vehicle battery, C new is the battery capacity of the vehicle battery at the initial stage.

[0080] In some embodiments, obtaining the predicted change value of the vehicle's own battery at the first preset number of cycles according to the mechanism model and big data includes: obtaining the predicted change value of the battery of other vehicles at the second preset number of cycles according to big data, where the second preset number of cycles is the number of cycles by which the battery of each vehicle among other vehicles has decayed within the first preset number of cycles; obtaining the predicted change value of the battery of the vehicle itself at the second preset number of cycles according to the mechanism model, and the predicted change value of the battery of the vehicle itself at the third preset number of cycles, where the third preset number of cycles is the number of cycles by which the battery of each vehicle among other vehicles has not decayed within the first preset number of cycles; first performing weighted calculation according to the predicted change value of the battery of other vehicles at the second preset number of cycles and the predicted change value of the battery of the vehicle itself at the second preset number of cycles, and adding the weighted calculation result to the predicted change value of the battery of the vehicle itself at the third preset number of cycles to obtain the predicted change value of the vehicle's own battery. The specific formula for obtaining the predicted change value of the vehicle's own battery is as follows:

[0081]

[0082] Wherein, when V SOHCi ≥V SOHC When When V SOHCi <V SOHC When ΔSOHC is the predicted change value of the vehicle's own battery, N is the number of vehicles meeting the screening conditions, P i Is the weight coefficient, ΔSOHCD i Is the predicted change value of the battery of other vehicles at the second preset number of cycles, ΔSOHCM is the predicted change value of the battery of the vehicle itself at the second preset number of cycles, and ΔSOHCMelse is the predicted change value of the battery of the vehicle itself at the third preset number of cycles.

[0083] That is to say, to obtain the predicted change value of the vehicle's own battery, it is necessary to separately obtain the predicted change value of the battery of other vehicles at the second preset number of cycles, the predicted change value of the battery of the vehicle itself at the second preset number of cycles, and the predicted change value of the battery of the vehicle itself at the third preset number of cycles. Among them, the second preset number of cycles is the number of cycles by which the battery of each vehicle among other vehicles has decayed within the first preset number of cycles, and the third preset number of cycles is the number of cycles by which the battery of each vehicle among other vehicles has not decayed within the first preset number of cycles. Assuming that the second preset number of cycles is from T0 times to T1 times, and the third preset number of cycles is from T1 times to T2 times, then the first preset number of cycles is from T0 times to T2 times.

[0084] Obtain the predicted change value of the battery life at the second preset number of cycles of other vehicles through big data, that is, obtain the predicted value of the battery life of other vehicles at the number of cycles T0 based on big data, and obtain the predicted value of the battery life of other vehicles at the number of cycles T1 based on big data. The difference between the two is the predicted change value of the battery life at the second preset number of cycles of other vehicles. The predicted value of the battery life of other vehicles at the number of cycles T0 and the predicted value of the battery life of other vehicles at the number of cycles T1 can be expressed as follows:

[0085]

[0086] Among them, SOHC iT0 is the predicted value of the battery life of other vehicles at the number of cycles T0 estimated based on historical big data, C nowiT0 is the current battery capacity of other vehicles at the number of cycles T0; C newi is the battery capacity at the initial stage of other vehicles, C featurenowiT0 is the battery capacity of the current characteristic voltage segment of other vehicles at the number of cycles T0; C featurenewi is the battery capacity of the initial stage characteristic voltage segment of other vehicles.

[0087]

[0088] Among them, SOHC iT1 is the predicted value of the battery life of other vehicles at the number of cycles T1 estimated based on historical big data, C nowiT1 is the current battery capacity of other vehicles at the number of cycles T1; C newi is the battery capacity at the initial stage of other vehicles, C featurenowiT1 is the battery capacity of the current characteristic voltage segment of other vehicles at the number of cycles T1; C featurenewi is the battery capacity of the initial stage characteristic voltage segment of other vehicles.

[0089] According to the predicted value of the battery life of other vehicles at the number of cycles T0 and the predicted value of the battery life of other vehicles at the number of cycles T1, the predicted change value of the battery life at the second preset number of cycles of other vehicles can be obtained. The specific formula is:

[0090] ΔSOHCD i =SOHC iT0 -SOHC iT1 ,

[0091] Obtain the predicted change value of the battery life when the vehicle itself reaches the second preset number of cycles according to the mechanism model, that is, obtain the predicted value of the vehicle's own battery life at the T0 number of cycles based on the mechanism model, and the predicted value of the vehicle's own battery life at the T1 number of cycles based on big data. The difference between the two is the predicted change value of the battery life when the vehicle itself reaches the second preset number of cycles. The predicted value of the vehicle's own battery life at the T0 number of cycles and the predicted value of the vehicle's own battery life at the T1 number of cycles can be expressed as follows respectively:

[0092]

[0093] Among them, SOHCJT0 is the predicted value of the vehicle's own battery life at the T0 number of cycles estimated based on the mechanism model, and Q loss，cycT0 is the battery capacity loss of the vehicle's own battery at the T0 number of cycles, and C new is the battery capacity of the vehicle's battery at the initial stage.

[0094]

[0095] Among them, Q loss，cycT0 is the battery capacity loss of the vehicle's battery at the T0 number of cycles, F is the Faraday constant, and t cyc is the time for one cycle, is the lithium ion loss, and T0 is the number of cycles.

[0096]

[0097] Among them, SOHCJT1 is the predicted value of the vehicle's own battery life at the T1 number of cycles estimated based on the mechanism model, and Q loss，cycT1 is the battery capacity loss of the vehicle's own battery at the T1 number of cycles, and C new is the battery capacity of the vehicle's battery at the initial stage.

[0098]

[0099] Among them, Q loss，cycT1 is the battery capacity loss of the vehicle's battery at the T1 number of cycles, F is the Faraday constant, and t cyc is the time for one cycle, is the lithium ion loss, and T1 is the number of cycles.

[0100] According to the predicted value of the vehicle's own battery life at the T0 number of cycles and the predicted value of the vehicle's own battery life at the T1 number of cycles, the predicted change value of the battery life when the vehicle itself reaches the second preset number of cycles can be obtained. The specific formula is as follows:

[0101] ΔSOHCM = SOHCJT0 - SOHCJT1,

[0102] Therefore, a weighted calculation is performed on the battery life prediction change value at the second preset cycle number of other vehicles and the battery life prediction change value at the second preset cycle number of the vehicle itself.

[0103] Obtain the battery life prediction change value at the third preset cycle number of the vehicle itself according to the mechanism model, that is, obtain the battery life prediction value of the vehicle itself at the T1 cycle number based on the mechanism model, and the battery life prediction value of the vehicle itself at the T2 cycle number based on big data. The difference between the two is the battery life prediction change value at the second preset cycle number of the vehicle itself. The battery life prediction value of the vehicle itself at the T1 cycle number has been obtained through the above formula. Therefore, only the battery life prediction value of the vehicle itself at the T2 cycle number needs to be obtained, and the specific expression is as follows:

[0104]

[0105] Among them, SOHCJT2 is the battery life prediction value of the vehicle itself at the T2 cycle number estimated based on the mechanism model, Q loss，cycT2 is the battery capacity loss of the vehicle itself at the T2 cycle number, C new is the initial battery capacity of the vehicle.

[0106]

[0107] Among them, Q loss，cycT2 is the battery capacity loss of the vehicle at the T2 cycle number, F is the Faraday constant, t cyc is the time for 1 cycle, is the lithium ion loss, and T2 is the cycle number.

[0108] According to the battery life prediction value of the vehicle itself at the T1 cycle number and the battery life prediction value of the vehicle itself at the T2 cycle number, the battery life prediction change value at the third preset cycle number of the vehicle itself can be obtained. The specific formula is as follows:

[0109] ΔSOHCMelse = SOHCJT1 - SOHCJT2,

[0110] Therefore, after performing a weighted calculation on the battery life prediction change value at the second preset cycle number of other vehicles and the battery life prediction change value at the second preset cycle number of the vehicle itself, add it to the battery life prediction change value at the third preset cycle number of the vehicle itself to obtain the battery life prediction change value of the vehicle itself.

[0111] Step S203: Perform an addition calculation according to the second life prediction value of the vehicle's own battery and the battery life prediction change value of the vehicle's own battery to obtain a second estimation result.

[0112] That is, after obtaining the second life prediction value of the vehicle's own battery and the life prediction change value of the vehicle's own battery at the first preset number of cycles, adding the second life prediction value of the vehicle's own battery to the life prediction change value of the vehicle's own battery can obtain the second estimation result. The specific formula of the second estimation result is as follows:

[0113] SOHC2 = SOHCself2 + ΔSOHC,

[0114] where SOHC2 is the second estimation result, SOHCself2 is the second life prediction value of the vehicle's own battery, and ΔSOHC is the life prediction change value of the vehicle's own battery.

[0115] Step S105: Predict the battery life using the corrected empirical model, and during the prediction process, correct the parameters of the corrected empirical model again according to the second estimation result to finally obtain the battery life prediction result.

[0116] Specifically, during the battery life prediction process, use the empirical model after the first parameter correction to predict the battery life, and use the estimated battery capacity state based on the mechanism model and big data obtained in step S104 above, that is, the second estimation result, as an input quantity to correct the parameters of the initial empirical model that has been corrected again, and at the same time output the final battery life prediction result.

[0117] In some embodiments, correcting the parameters of the corrected empirical model again according to the second estimation result includes: using the second learning result output by the corrected empirical model as a priori estimation, and using the second estimation result as a posteriori correction, and using the Kalman filter algorithm to correct the parameters of the corrected empirical model again.

[0118] That is, when performing the second parameter correction on the empirical model, as Figure 2 shown, use the battery life prediction value obtained during the first parameter correction as the second learning result, and use this second learning result as a priori estimation, and use the above second estimation result, that is, the estimated battery capacity state based on the mechanism model and big data, as a posteriori correction. Use the Kalman filter algorithm to correct the parameters of the empirical model after the first correction again according to the input second learning result and the second estimation result, and use the empirical model with the second parameter correction during the battery life prediction process to improve the battery life prediction accuracy once again.

[0119] In summary, according to the mutual learning prediction method for vehicle battery life in the embodiments of the present invention, an empirical model is determined, and the historical data trajectory of the vehicle's own battery is obtained, as well as the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery. The empirical model is used to perform mutual learning on the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries, estimate the battery capacity state based on big data to obtain a first estimation result, and during the mutual learning process, correct the parameters of the empirical model according to the first estimation result, determine a mechanism model, and estimate the battery capacity state according to the mechanism model and big data to obtain a second estimation result. The corrected empirical model is used to predict the battery life, and during the prediction process, the parameters of the corrected empirical model are corrected again according to the second estimation result, and finally the battery life prediction result is obtained. Thus, based on big data and the determined mechanism model, the parameters of the determined empirical model are corrected twice successively, so that a more accurate battery life prediction model can be constructed, and the accuracy of battery life prediction is improved.

[0120] An embodiment of the present invention also provides a computer-readable storage medium, on which a mutual learning prediction program for vehicle battery life is stored. When the mutual learning prediction program for vehicle battery life is executed by a processor, the above-mentioned mutual learning prediction method for vehicle battery life is implemented.

[0121] According to the computer-readable storage medium in the embodiments of the present invention, through the above-mentioned mutual learning prediction method for vehicle battery life, based on big data and the determined mechanism model, the parameters of the determined empirical model are corrected twice successively, so that a more accurate battery life prediction model can be constructed, and the accuracy of battery life prediction is improved.

[0122] An embodiment of the present invention also provides a cloud server, including a memory, a processor, and a mutual learning prediction program for vehicle battery life stored on the memory and executable on the processor. When the processor executes the mutual learning prediction program for vehicle battery life, the above-mentioned mutual learning prediction method for vehicle battery life is implemented.

[0123] According to the cloud server in the embodiments of the present invention, through the above-mentioned mutual learning prediction method for vehicle battery life, based on big data and the determined mechanism model, the parameters of the determined empirical model are corrected twice successively, so that a more accurate battery life prediction model can be constructed, and the accuracy of battery life prediction is improved.

[0124] Figure 4 It is a schematic structural diagram of a mutual learning prediction device for vehicle battery life according to an embodiment of the present invention. As Figure 4As shown in the figure, the mutual learning prediction device 100 for the vehicle battery life includes: a first determination module 110, a first estimation module 120, a first correction module 130, a second determination module 140, a second estimation module 150, and a prediction module 160.

[0125] Among them, the first determination module 110 is used to determine the empirical model, obtain the historical data trajectory of the vehicle's own battery, and obtain the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery, and perform mutual learning on the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries using the empirical model; the first estimation module 120 estimates the battery capacity state based on big data to obtain a first estimation result; the first correction module 130 is used to correct the parameters of the empirical model according to the first estimation result during the mutual learning process; the second determination module 140 is used to determine the mechanism model; the second estimation module 150 is used to estimate the battery capacity state according to the mechanism model and big data to obtain a second estimation result; the prediction module 160 is used to predict the battery life using the corrected empirical model, and during the prediction process, correct the parameters of the corrected empirical model again according to the second estimation result to finally obtain the battery life prediction result.

[0126] In some embodiments, the first determination module 110 is specifically configured to: when the battery type of other vehicles is the same as that of the vehicle's own battery, if the battery cycle times of other vehicles are greater than those of the vehicle's own battery, the battery life prediction values of other vehicles are less than the life prediction values of the vehicle's own battery, and the battery attenuation rates of other vehicles and the vehicle's own battery meet the preset conditions, then determine that the historical data trajectory of other vehicle batteries is adapted to the historical data trajectory of the vehicle's own battery.

[0127] In some embodiments, when the following relational expression is satisfied between the battery attenuation rates of other vehicles and the vehicle's own battery, it is determined that the battery attenuation rates of other vehicles and the vehicle's own battery meet the preset conditions: where, V SOHCi is the battery attenuation rate of other vehicles, and V SOHC is the battery attenuation rate of the vehicle's own battery.

[0128] In some embodiments, the first correction module 130 is specifically configured to: use the first learning result as the prior estimate, and use the first estimation result as the posterior correction, and use the Kalman filter algorithm to correct the parameters of the empirical model, where the first learning result is obtained by the empirical model learning the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries.

[0129] In some embodiments, the prediction module 160 is specifically configured to: use the second learning result output by the corrected empirical model as a prior estimate, and use the second estimation result as a posterior correction, and then use the Kalman filtering algorithm to correct the parameters of the corrected empirical model again.

[0130] In some embodiments, the first estimation module 120 is specifically configured to: obtain the first life prediction value of the vehicle's own battery and the life prediction values of other vehicles' batteries at the same number of cycles; perform weighted calculation based on the first life prediction value of the vehicle's own battery and the life prediction values of other vehicles' batteries to obtain a first estimation result.

[0131] In some embodiments, the first estimation result is calculated according to the following formula:

[0132]

[0133] where, when V SOHCi ≥V SOHC then when V SOHCi <V SOHC then SOHC1 is the first estimation result, N is the number of vehicles meeting the screening conditions, P i is the weight coefficient, SOHC i is the life prediction value of other vehicles' batteries, and SOHCself1 is the first life prediction value of the vehicle's own battery.

[0134] In some embodiments, the second estimation module 150 is specifically configured to: obtain the second life prediction value of the vehicle's own battery at the first preset number of cycles; obtain the life prediction change value of the vehicle's own battery at the first preset number of cycles according to the mechanism model and big data; perform addition calculation based on the second life prediction value of the vehicle's own battery and the life prediction change value of the vehicle's own battery to obtain a second estimation result.

[0135] In some embodiments, the second estimation result is calculated according to the following formula:

[0136] SOHC2 = SOHCself2 + ΔSOHC,

[0137] where, SOHC2 is the second estimation result, SOHCself2 is the second life prediction value of the vehicle's own battery, and ΔSOHC is the life prediction change value of the vehicle's own battery.

[0138] In some embodiments, the second estimation module 150 is specifically configured to: obtain the predicted change value of the battery life at the second preset number of cycles of other vehicles according to big data, where the second preset number of cycles is the number of cycles by which the batteries of each vehicle among other vehicles have decayed within the first preset number of cycles; obtain the predicted change value of the battery life at the second preset number of cycles of the vehicle itself and the predicted change value of the battery life at the third preset number of cycles of the vehicle itself according to the mechanism model, where the third preset number of cycles is the number of cycles by which the batteries of each vehicle among other vehicles have not decayed within the first preset number of cycles; first perform weighted calculation according to the predicted change value of the battery life at the second preset number of cycles of other vehicles and the predicted change value of the battery life at the second preset number of cycles of the vehicle itself, and add the weighted calculation result to the predicted change value of the battery life at the third preset number of cycles of the vehicle itself to obtain the predicted change value of the battery life of the vehicle itself.

[0139] In some embodiments, the predicted change value of the battery life of the vehicle itself is calculated according to the following formula:

[0140]

[0141] Where, when V SOHCi ≥V SOHC At this time, When V SOHCi <V SOHC At this time, ΔSOHC is the predicted change value of the battery life of the vehicle itself, N is the number of vehicles that meet the screening conditions, P i Is the weight coefficient, ΔSOHCD i Is the predicted change value of the battery life at the second preset number of cycles of other vehicles, ΔSOHCM is the predicted change value of the battery life at the second preset number of cycles of the vehicle itself, and ΔSOHCMelse is the predicted change value of the battery life at the third preset number of cycles of the vehicle itself.

[0142] It should be noted that for the description of the mutual learning prediction device for vehicle battery life in this application, please refer to the description of the mutual learning prediction method for vehicle battery life in this application, and details are not elaborated here.

[0143] The mutual learning prediction device for the vehicle battery life according to the embodiment of the present invention determines an empirical model through a first determination module, obtains the historical data trajectory of the vehicle's own battery, and obtains the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery, and performs mutual learning on the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries by using the empirical model. Through a first estimation module, the battery capacity state is estimated based on big data to obtain a first estimation result. During the mutual learning process, a first correction module corrects the parameters of the empirical model according to the first estimation result. A second determination module determines a mechanism model, and a second estimation module estimates the battery capacity state according to the mechanism model and big data to obtain a second estimation result. A prediction module uses the corrected empirical model to predict the battery life, and during the prediction process, the parameters of the corrected empirical model are corrected again according to the second estimation result, and finally the battery life prediction result is obtained. Thus, based on big data and the determined mechanism model, the parameters of the determined empirical model are corrected twice successively, so that a more accurate battery life prediction model can be constructed, and the accuracy of battery life prediction is improved.

[0144] It should be noted that the logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus or device and execute the instructions), or in combination with these instruction execution systems, apparatus or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion having one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then stored in a computer memory.

[0145] It should be understood that each part of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits with logic gate circuits for implementing logical functions on data signals, application specific integrated circuits with appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0146] In the description of this specification, the descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", 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 can be combined in any one or more embodiments or examples in a suitable manner.

[0147] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In the description of the present invention, the meaning of "a plurality" is at least two, such as two, three, etc., unless otherwise specifically and clearly defined.

[0148] In the present invention, unless otherwise clearly specified and defined, terms such as "install", "connect", "connection", "fix", etc. should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the internal connection or the interaction relationship between two components, unless otherwise clearly defined. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0149] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A mutual learning prediction method for vehicle battery life, characterized in that, Including: Determine an empirical model, obtain the historical data trajectory of the vehicle's own battery, and obtain the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery; Use the empirical model to perform mutual learning on the historical data trajectory of the vehicle's own battery and the historical data trajectory of other vehicle batteries; Estimate the battery capacity state based on big data to obtain a first estimation result, and during the mutual learning process, correct the parameters of the empirical model according to the first estimation result; Determine a mechanism model, and estimate the battery capacity state according to the mechanism model and big data to obtain a second estimation result; Use the corrected empirical model to predict the battery life, and during the prediction process, correct the parameters of the corrected empirical model again according to the second estimation result to finally obtain the battery life prediction result; Estimate the battery capacity state based on big data to obtain a first estimation result, including: Obtain the first life prediction value of the vehicle's own battery and the life prediction value of other vehicle batteries at the same number of cycles; Perform weighted calculation according to the first life prediction value of the vehicle's own battery and the life prediction value of other vehicle batteries to obtain the first estimation result; Calculate the first estimation result according to the following formula: Among them, at V SoHCi ≥ V SOHC When At V SOHCi < V SOHC When SOHC1 is the first estimation result, N is the number of vehicles meeting the screening conditions, P i Is the weight coefficient, SOHC i Is the predicted life of the batteries of the other vehicles, and SOHCself1 is the first predicted life of the vehicle's own battery; Estimate the battery capacity state according to the mechanism model and big data to obtain a second estimation result, including: Obtain the second life prediction value of the vehicle's own battery at the first preset number of cycles; Obtain the life prediction change value of the vehicle's own battery at the first preset number of cycles according to the mechanism model and big data; Perform addition calculation according to the second life prediction value of the vehicle's own battery and the life prediction change value of the vehicle's own battery to obtain the second estimation result; Calculate the second estimation result according to the following formula: SOHC2 = SOHCself2 + ΔSOHC, where SOHC2 is the second estimation result, SOHCself2 is the second life prediction value of the vehicle's own battery, and ΔSOHC is the life prediction change value of the vehicle's own battery.

2. The mutual learning prediction method for battery life according to claim 1, characterized in that, Obtain the historical data trajectory of other vehicle batteries adapted to the vehicle's own battery, including: When the battery type of other vehicles is the same as that of the vehicle's own battery, if the number of battery cycles of other vehicles is greater than the number of cycles of the vehicle's own battery, the life prediction value of other vehicle batteries is less than the life prediction value of the vehicle's own battery, and the battery decay rate of other vehicles and the battery decay rate of the vehicle's own battery meet the preset conditions, then determine that the historical data trajectory of other vehicle batteries is adapted to the historical data trajectory of the vehicle's own battery.

3. The mutual learning prediction method for battery life according to claim 2, characterized in that, When the following relational expression is satisfied between the battery decay rate of other vehicles and the battery decay rate of the vehicle's own battery, determine that the battery decay rate of other vehicles and the battery decay rate of the vehicle's own battery meet the preset conditions: Among them, V SOHCi is the battery attenuation rate of the other vehicle, and V SOHC is the battery attenuation rate of the vehicle itself.

4. The mutual learning prediction method for battery life according to claim 1, characterized in that, Correct the parameters of the empirical model according to the first estimation result, including: Take the first learning result as the prior estimate, and take the first estimation result as the posterior correction, and use the Kalman filter algorithm to correct the parameters of the empirical model, where the first learning result is obtained by learning the historical data trajectory of the vehicle's own battery and the historical data trajectory of the batteries of other vehicles according to the empirical model.

5. The mutual learning prediction method for battery life according to claim 4, characterized in that, Correct the parameters of the corrected empirical model again according to the second estimation result, including: Take the second learning result output by the corrected empirical model as the prior estimate, and take the second estimation result as the posterior correction, and use the Kalman filter algorithm to correct the parameters of the corrected empirical model again.

6. The mutual learning prediction method for battery life according to claim 1, characterized in that, The obtaining of the predicted change value of the life of the vehicle's own battery at the first preset number of cycles according to the mechanism model and big data includes: Obtain the predicted change value of the battery life of other vehicles at the second preset number of cycles according to big data, where the second preset number of cycles is the number of cycles by which the batteries of each vehicle among the other vehicles have decayed within the first preset number of cycles; Obtain the predicted change value of the battery life of the vehicle's own at the second preset number of cycles according to the mechanism model, and the predicted change value of the battery life of the vehicle's own at the third preset number of cycles, where the third preset number of cycles is the number of cycles by which the batteries of each vehicle among the other vehicles have not decayed within the first preset number of cycles; First perform weighted calculation according to the predicted change value of the battery life of other vehicles at the second preset number of cycles and the predicted change value of the battery life of the vehicle's own at the second preset number of cycles, and add the weighted calculation result to the predicted change value of the battery life of the vehicle's own at the third preset number of cycles to obtain the predicted change value of the battery life of the vehicle's own.

7. The mutual learning prediction method for battery life according to claim 6, wherein, Calculate the predicted change value of the battery life of the vehicle's own according to the following formula: Among them, at V SOHCi ≥V SOHC When At V SOHCi <V SOHC When ΔSOHC is the predicted change value of the life of the vehicle's own battery, N is the number of vehicles meeting the screening conditions, P i is the weight coefficient, ΔSOHCD i is the predicted change value of the battery life at the second preset cycle number of the other vehicles, ΔSOHCM is the predicted change value of the battery life at the second preset cycle number of the vehicle itself, and ΔSOHCMelse is the predicted change value of the battery life at the third preset cycle number of the vehicle itself.

8. A computer-readable storage medium, wherein, A mutual learning prediction program for the life of the vehicle battery is stored thereon. When the mutual learning prediction program for the life of the vehicle battery is executed by a processor, the mutual learning prediction method for the life of the vehicle battery according to any one of claims 1-7 is implemented.

9. A cloud server, wherein, It includes a memory, a processor, and a mutual learning prediction program for the life of the vehicle battery stored on the memory and operable on the processor. When the processor executes the mutual learning prediction program for the life of the vehicle battery, the mutual learning prediction method for the life of the vehicle battery according to any one of claims 1-7 is implemented.

10. A mutual learning prediction device for vehicle battery life, wherein, It includes: A first determination module for determining an empirical model, obtaining the historical data trajectory of the vehicle's own battery, obtaining the historical data trajectory of the batteries of other vehicles adapted to the vehicle's own battery, and performing mutual learning on the historical data trajectory of the vehicle's own battery and the historical data trajectory of the batteries of other vehicles using the empirical model; A first estimation module for estimating the battery capacity state based on big data to obtain a first estimation result; A first correction module for correcting the parameters of the empirical model according to the first estimation result during the mutual learning process; A second determination module for determining a mechanism model; A second estimation module for estimating the battery capacity state according to the mechanism model and big data to obtain a second estimation result; A prediction module, which is used to predict the battery life by using the corrected empirical model, and during the prediction process, correct the parameters of the corrected empirical model again according to the second estimation result, and finally obtain the battery life prediction result; Estimate the battery capacity state based on big data to obtain a first estimation result, including: Obtain the first life prediction value of the vehicle's own battery and the life prediction values of the batteries of other vehicles at the same number of cycles; Perform weighted calculation according to the first life prediction value of the vehicle's own battery and the life prediction values of the batteries of other vehicles to obtain the first estimation result; Calculate the first estimation result according to the following formula: Among them, at V SOHCi ≥V SOHC When At V SOHCi <V SOHC When SOHC1 is the first estimation result, N is the number of vehicles meeting the screening conditions, P i is the weight coefficient, SOHC i is the predicted life of the batteries of the other vehicles, and SOHCself1 is the first predicted life of the vehicle's own battery; Estimate the battery capacity state according to the mechanism model and big data to obtain a second estimation result, including: Obtain the second life prediction value of the vehicle's own battery at the first preset number of cycles; Obtain the life prediction change value of the vehicle's own battery at the first preset number of cycles according to the mechanism model and big data; Perform addition calculation according to the second life prediction value of the vehicle's own battery and the life prediction change value of the vehicle's own battery to obtain the second estimation result; Calculate the second estimation result according to the following formula: SOHC2 = SOHCself2 + ΔSOHC, where SOHC2 is the second estimation result, SOHCself2 is the second life prediction value of the vehicle's own battery, and ΔSOHC is the life prediction change value of the vehicle's own battery.

Images