Power battery capacity prediction method based on electric vehicle capacity mileage data

Abstract

The invention discloses a power battery capacity prediction method based on electric vehicle capacity mileage data, and the method comprises the steps: obtaining the mileage to be predicted of a target vehicle, and determining the historical mileage capacity data of the target vehicle and the overall mileage capacity data of a reference vehicle; determining a first capacity according to the mileage to be predicted and the overall mileage capacity data; based on a multi-output model, determining a second capacity according to the historical mileage capacity data, the to-be-predicted mileage and the overall mileage capacity data; and determining the predicted capacity corresponding to the mileage to be predicted according to the first capacity and the second capacity. Based on the second capacity obtained by the multi-output model, the internal correlation between the aging tracks of the power batteries in the reference vehicle and the target vehicle is established so as to predict the capacity of the target vehicle, the predicted capacity is obtained by combining the first capacity and the second capacity, and the first capacity is used for correcting the second capacity, so that the reliability of the predicted capacity is ensured; and the method is relatively high in universality.

Description

Technical Field

[0001] The present application relates to the technical field of electric vehicles, and in particular to a method for predicting the capacity of a power battery based on the capacity mileage data of an electric vehicle. Background Art

[0002] As the number of times electric vehicles are charged and discharged increases, irreversible chemical reactions occur inside the battery, and its performance will gradually decline. The most obvious is the decrease in capacity and the increase in internal resistance, which makes it difficult to continue to meet the use needs and greatly increases the safety risks of the battery. In the prior art, the data complexity in the actual application scenarios of electric vehicles is high. By predicting the number of charge and discharge cycles or the remaining use time of the electric vehicle power battery as the battery failure point, the reliability of the power battery life or capacity prediction is not high, and it is difficult to have universality, and it is not generally reliable for most electric vehicles. Summary of the invention

[0003] The present application provides a method for predicting the capacity of a power battery based on the capacity mileage data of an electric vehicle, so as to solve the technical problem in the related art that the reliability of the life or capacity prediction of the power battery is not high and it is difficult to have universality.

[0004] The first embodiment of the present application provides a method for predicting the capacity of a power battery based on the capacity mileage data of an electric vehicle, comprising the steps of:

[0005] Acquire the mileage to be predicted of the target vehicle, and determine the historical mileage capacity data of the target vehicle and the overall mileage capacity data of the reference vehicle corresponding to the target vehicle; wherein the mileage to be predicted is greater than the maximum mileage in the historical mileage capacity data, and the mileage to be predicted is less than the maximum mileage in the overall mileage capacity data;

[0006] Determining a first capacity according to the mileage to be predicted and the overall mileage capacity data;

[0007] Determining a second capacity based on the multi-output model according to the historical mileage capacity data, the mileage to be predicted, and the overall mileage capacity data;

[0008] A predicted capacity corresponding to the mileage to be predicted is determined according to the first capacity and the second capacity.

[0009] According to the above technical means, the second capacity obtained based on the multi-output model does not require too much overall mileage capacity data of the reference vehicle, and the intrinsic correlation between the aging trajectories of the power batteries in the reference vehicle and the target vehicle can be established to predict the capacity of the target vehicle. The first capacity obtained based on the overall mileage capacity data is closer to the actual capacity of the electric vehicle in the actual application scenario. The predicted capacity is obtained by combining the first capacity and the second capacity, and the second capacity is corrected by using the first capacity, which ensures the reliability of the predicted capacity. It is suitable for target vehicles with different user habits and different working conditions, and has strong versatility.

[0010] Optionally, determining the historical mileage capacity data of the target vehicle and the overall mileage capacity data of a reference vehicle corresponding to the target vehicle comprises:

[0011] Acquire vehicle data of a target vehicle and vehicle data of several candidate vehicles; wherein the vehicle data includes several mileage intervals and the capacity corresponding to each mileage in the mileage intervals;

[0012] Based on the capacity decay model, according to the vehicle data of the target vehicle, determine the historical mileage capacity data of the target vehicle;

[0013] Determining candidate mileage capacity data of the candidate vehicle based on the capacity decay model and according to the vehicle data of the candidate vehicle;

[0014] Based on the historical mileage capacity data, the overall mileage capacity data is determined from a plurality of candidate mileage capacity data, and the candidate vehicles corresponding to the overall mileage capacity data are used as reference vehicles corresponding to the target vehicle.

[0015] According to the above technical means, when the capacity attenuation trajectory of the historical mileage capacity data is similar or close to the capacity attenuation trajectory of the candidate mileage capacity data, the candidate mileage capacity data is used as the overall mileage capacity data, and the candidate vehicle corresponding to the candidate mileage capacity data is used as the reference vehicle, thereby facilitating accurate prediction of the capacity attenuation of the target vehicle and obtaining an accurate first capacity.

[0016] Optionally, the historical mileage capacity data includes: a number of historically reconstructed mileage intervals and historically reconstructed capacities corresponding to the historically reconstructed mileage intervals;

[0017] The method of determining the historical mileage capacity data of the target vehicle based on the capacity decay model and according to the vehicle data of the target vehicle includes:

[0018] Based on the capacity decay model, for each mileage interval of the target vehicle, the mileage and the corresponding capacity in the mileage interval are fitted to obtain a target reconstruction curve;

[0019] Based on the target reconstruction curve, a historical reconstruction mileage interval and a historical reconstruction capacity corresponding to the historical reconstruction mileage interval are determined.

[0020] According to the above technical means, a piecewise fitting method is adopted to fit the capacity in each mileage interval. After all mileage intervals are fitted, a target reconstruction curve is obtained, and the target reconstruction curve is more accurate.

[0021] Optionally, the candidate mileage capacity data includes: a plurality of candidate reconstructed mileage intervals and candidate reconstructed capacities corresponding to the candidate reconstructed mileage intervals;

[0022] The determining, based on the capacity decay model and according to the vehicle data of the candidate vehicle, the candidate mileage capacity data of the candidate vehicle comprises:

[0023] Based on the capacity decay model, for each mileage interval of the candidate vehicle, the mileage in the mileage interval and the corresponding capacity are fitted to obtain a candidate reconstruction curve;

[0024] Based on the candidate reconstruction curve, a candidate reconstruction mileage interval and a candidate reconstruction capacity corresponding to the candidate reconstruction mileage interval are determined.

[0025] According to the above technical means, a piecewise fitting method is adopted to fit the capacity in each mileage interval. After all mileage intervals are fitted, a candidate reconstruction curve is obtained, and the accuracy of the candidate reconstruction curve is relatively high.

[0026] Optionally, the capacity attenuation model is:

[0027] y=ae bx +cx 2 +d;

[0028] Among them, y represents capacity, e represents natural constant, a, b, c, d represent model parameters, and x represents mileage.

[0029] According to the above technical means, the capacity decay model uses exponential and quadratic polynomials to fit the mileage and capacity. Although different mileage intervals are based on the same capacity decay model, the model parameters obtained by fitting are different, which makes the fitting accuracy better.

[0030] Optionally, the overall mileage capacity data includes: a plurality of overall reconstructed mileage intervals and overall reconstructed capacities corresponding to the overall reconstructed mileage intervals;

[0031] The determining of the overall mileage capacity data from a plurality of candidate mileage capacity data based on the historical mileage capacity data comprises:

[0032] For each candidate vehicle, determine the cumulative distance cost of the candidate vehicle; wherein the cumulative distance cost is the cumulative value of the distances between a number of the historical reconstructed mileage intervals and their corresponding historical reconstructed capacities, and a number of the candidate reconstructed mileage intervals and their corresponding candidate reconstructed capacities;

[0033] The candidate reconstructed mileage interval corresponding to the minimum value of the cumulative distance cost of the candidate vehicles is used as the overall reconstructed mileage interval, and the candidate reconstructed capacity corresponding to the minimum value of the cumulative distance cost of the candidate vehicles is used as the overall reconstructed capacity.

[0034] According to the above technical means, the smaller the cumulative distance cost is, the smaller the DTW distance between the historical reconstructed mileage interval sequence and the candidate reconstructed capacity sequence is, and the closer the similarity between the historical reconstructed mileage interval sequence and the candidate reconstructed capacity sequence is. The candidate vehicle corresponding to the minimum value in the cumulative distance cost can be used as a reference vehicle.

[0035] Optionally, the cumulative distance cost is:

[0036] C(i,j)=D(i,j)+min(C(i-1,j),C(i,j-1),C(i-1,j-1));

[0037] C(0,0) <C(0,j)=C(i,0)=∞;

[0038] Among them, C(i,j) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-th candidate reconstruction mileage interval, D(i,j) represents the distance between the historical reconstruction capacity corresponding to the i-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-th candidate reconstruction mileage interval, C(i-1,j) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-1-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-1-th candidate reconstruction mileage interval, C(i,j-1) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-1-th candidate reconstruction mileage interval, C(i-1,j-1) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-1-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-1-th candidate reconstruction mileage interval, and min(·) represents the minimum value operation.

[0039] According to the above technical means, the cumulative distance costs C(n,m) calculated for different candidate vehicles are different. The candidate vehicle corresponding to the minimum value of all cumulative distance costs C(n,m) is used as the reference vehicle, which facilitates the rapid determination of the reference vehicle.

[0040] Optionally, determining the first capacity according to the mileage to be predicted and the overall mileage capacity data includes:

[0041] Determining the overall reconstructed mileage interval corresponding to the mileage to be predicted according to the mileage to be predicted and a plurality of the overall reconstructed mileage intervals;

[0042] The overall reconstructed capacity corresponding to the overall reconstructed mileage interval is used as the first capacity; or, according to the model parameters of the overall reconstructed mileage interval in the fitting process and the mileage to be predicted, the capacity corresponding to the mileage to be predicted is determined, and the capacity is used as the first capacity.

[0043] According to the above technical means, after determining the reference vehicle and the overall mileage capacity data of the reference vehicle, the corresponding overall reconstructed mileage interval can be determined according to the mileage to be predicted, and the overall reconstructed capacity corresponding to the overall reconstructed mileage interval is used as the first capacity. The obtained first capacity conforms to the actual application scenario of the reference vehicle.

[0044] Optionally, the determining the second capacity based on the multi-output model according to the historical mileage capacity data, the mileage to be predicted and the overall mileage capacity data includes:

[0045] Training a multi-output model according to the historically reconstructed mileage interval, the historically reconstructed capacity, the overall reconstructed mileage interval, and the overall reconstructed capacity to obtain a trained multi-output model;

[0046] The mileage to be predicted is input into the trained multi-output model, and the second capacity is output through the trained multi-output model.

[0047] According to the above technical means, a multi-output model is used to output the second capacity. The second capacity reflects the capacity data with a high probability for most vehicles, which is convenient for predicting application scenarios with different user habits and different working conditions, and improves the versatility of the prediction method.

[0048] Optionally, the multi-output model is:

[0049] y~GP(m(x),K M );

[0050]

[0051] K x =(k pq (xx′));

[0052]

[0053] Where y represents capacity, GP(·,·) represents Gaussian process, m(·) represents mean function, and KM represents the covariance function, K f represents the relationship matrix, represents the Kronecker product, K x represents the correlation matrix of mileage, k pq (xx′) represents K x elements in , σ f , ρ are hyperparameters, υ is a smoothing parameter, R v (·) represents the modified Bessel function, x and x′ represent the mileage, and Γ(·) represents the gamma function.

[0054] According to the above technical means, a multi-output model of multi-output Gaussian process regression is used, which is conducive to quickly calculating the second capacity.

[0055] Optionally, the predicted capacity is:

[0056] y pre =α*y1+(1-α)*y2;

[0057] Among them, y pre represents the predicted capacity, α represents the weight, y1 represents the first capacity, and y2 represents the second capacity.

[0058] According to the above technical means, combining the first capacity and the second capacity can make the predicted capacity more reliable.

[0059] Optionally, the historical mileage capacity data of the target vehicle includes an initial capacity;

[0060] The power battery capacity prediction method further comprises the steps of:

[0061] The remaining life of the power battery of the target vehicle is determined according to the initial capacity and the predicted capacity.

[0062] According to the above technical means, the remaining life of the power battery of the target vehicle can be determined according to the initial capacity and the predicted capacity, and the prediction of the remaining life of the power battery is more accurate.

[0063] The second aspect of the present application provides a power battery capacity prediction system based on electric vehicle capacity mileage data, including:

[0064] an acquisition module, used for acquiring the mileage to be predicted of the target vehicle, and determining the historical mileage capacity data of the target vehicle and the overall mileage capacity data of the reference vehicle corresponding to the target vehicle; wherein the time corresponding to the mileage to be predicted is later than the time corresponding to the historical mileage capacity data, and the mileage to be predicted is less than the maximum mileage in the overall mileage capacity data;

[0065] A first capacity module, configured to determine a first capacity according to the mileage to be predicted and the overall mileage capacity data;

[0066] A second capacity module, configured to determine a second capacity based on a multi-output model according to the historical mileage capacity data, the mileage to be predicted and the overall mileage capacity data;

[0067] A prediction module is used to determine the predicted capacity corresponding to the mileage to be predicted according to the first capacity and the second capacity.

[0068] Optionally, the acquisition module is specifically used to acquire vehicle data of a target vehicle and vehicle data of several candidate vehicles; wherein the vehicle data includes several mileage intervals and the capacity corresponding to each mileage within the mileage intervals; based on a capacity decay model, the historical mileage capacity data of the target vehicle is determined according to the vehicle data of the target vehicle; based on a capacity decay model, the candidate mileage capacity data of the candidate vehicle is determined according to the vehicle data of the candidate vehicle; based on the historical mileage capacity data, the overall mileage capacity data is determined from several of the candidate mileage capacity data, and the candidate vehicle corresponding to the overall mileage capacity data is used as a reference vehicle corresponding to the target vehicle.

[0069] Optionally, the historical mileage capacity data includes: a number of historically reconstructed mileage intervals and the historically reconstructed capacities corresponding to the historically reconstructed mileage intervals; the acquisition module is specifically used to fit the mileage and the corresponding capacity within each mileage interval of the target vehicle based on a capacity attenuation model to obtain a target reconstruction curve; based on the target reconstruction curve, determine the historically reconstructed mileage interval and the historically reconstructed capacity corresponding to the historically reconstructed mileage interval.

[0070] Optionally, the candidate mileage capacity data includes: a number of candidate reconstructed mileage intervals and candidate reconstructed capacities corresponding to the candidate reconstructed mileage intervals; the acquisition module is specifically used to fit the mileage and the corresponding capacity within each mileage interval of the candidate vehicle based on a capacity decay model to obtain a candidate reconstruction curve; based on the candidate reconstruction curve, determine the candidate reconstructed mileage interval and the candidate reconstruction capacity corresponding to the candidate reconstructed mileage interval.

[0071] Optionally, the capacity attenuation model is:

[0072] y=ae bx +cx 2 +d;

[0073] Among them, y represents capacity, e represents natural constant, a, b, c, d represent model parameters, and x represents mileage.

[0074] Optionally, the overall mileage capacity data includes: a number of overall reconstructed mileage intervals and the overall reconstructed capacities corresponding to the overall reconstructed mileage intervals; the acquisition module is specifically used to determine the cumulative distance cost of the candidate vehicle for each candidate vehicle; wherein the cumulative distance cost is the cumulative value of the distances between a number of the historical reconstructed mileage intervals and their respective corresponding historical reconstruction capacities, and a number of the candidate reconstructed mileage intervals and their respective corresponding candidate reconstruction capacities; the candidate reconstructed mileage interval corresponding to the minimum value of the cumulative distance cost of the candidate vehicle is taken as the overall reconstructed mileage interval, and the candidate reconstruction capacity corresponding to the minimum value of the cumulative distance cost of the candidate vehicle is taken as the overall reconstruction capacity.

[0075] Optionally, the cumulative distance cost is:

[0076] C(i,j)=D(i,j)+min(C(i-1,j),C(i,j-1),C(i-1,j-1));

[0077] C(0,0) <C(0,j)=C(i,0)=∞;

[0078] Among them, C(i,j) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-th candidate reconstruction mileage interval, D(i,j) represents the distance between the historical reconstruction capacity corresponding to the i-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-th candidate reconstruction mileage interval, C(i-1,j) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-1-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-1-th candidate reconstruction mileage interval, C(i,j-1) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-1-th candidate reconstruction mileage interval, C(i-1,j-1) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-1-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-1-th candidate reconstruction mileage interval, and min(·) represents the minimum value operation.

[0079] Optionally, the first capacity module is specifically used to determine the overall reconstructed mileage interval corresponding to the mileage to be predicted based on the mileage to be predicted and several of the overall reconstructed mileage intervals; use the overall reconstructed capacity corresponding to the overall reconstructed mileage interval as the first capacity; or, determine the capacity corresponding to the mileage to be predicted based on the model parameters of the overall reconstructed mileage interval during the fitting process and the mileage to be predicted, and use the capacity as the first capacity.

[0080] Optionally, the second capacity module is specifically used to train the multi-output model according to the historical reconstructed mileage interval, the historical reconstructed capacity, the overall reconstructed mileage interval and the overall reconstructed capacity to obtain a trained multi-output model; input the mileage to be predicted into the trained multi-output model, and output the second capacity through the trained multi-output model.

[0081] Optionally, the multi-output model is:

[0082] y~GP(m(x),K M );

[0083]

[0084] K x =(k pq (xx′));

[0085]

[0086] Where y represents capacity, GP(·,·) represents Gaussian process, m(·) represents mean function, and K M represents the covariance function, K f represents the relationship matrix, represents the Kronecker product, K x represents the correlation matrix of mileage, k pq (xx′) represents K x elements in , σ f , ρ are hyperparameters, υ is a smoothing parameter, R v (·) represents the modified Bessel function, x and x′ represent the mileage, and Γ(·) represents the gamma function.

[0087] Optionally, the predicted capacity is:

[0088] y pre =α*y1+(1-α)*y2;

[0089] Among them, y pre represents the predicted capacity, α represents the weight, y1 represents the first capacity, and y2 represents the second capacity.

[0090] Optionally, the historical mileage capacity data of the target vehicle includes an initial capacity; the power battery capacity prediction system based on the electric vehicle capacity mileage data further includes:

[0091] The remaining life module is used to determine the remaining life of the power battery of the target vehicle according to the initial capacity and the predicted capacity.

[0092] The third aspect of the present application provides a vehicle-mounted terminal, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a power battery capacity prediction method based on electric vehicle capacity mileage data as described in the above embodiment.

[0093] A fourth aspect of the present application provides a vehicle, comprising: a power battery capacity prediction system based on electric vehicle capacity mileage data as described in the above embodiments, or a vehicle-mounted terminal as described in the above embodiments.

[0094] The beneficial effects of the present application are as follows: the second capacity obtained based on the multi-output model does not require too much overall mileage capacity data of the reference vehicle, and the intrinsic correlation between the aging trajectory of the power batteries in the reference vehicle and the target vehicle can be established to predict the capacity of the target vehicle, while the first capacity obtained based on the overall mileage capacity data is closer to the actual capacity of the electric vehicle in the actual application scenario. The predicted capacity is obtained by combining the first capacity and the second capacity, and the second capacity is corrected by using the first capacity, which ensures the reliability of the predicted capacity, and is suitable for target vehicles with different user habits and different working conditions, and has strong versatility. BRIEF DESCRIPTION OF THE DRAWINGS

[0095] Figure 1 is a flow chart of a method for predicting power battery capacity based on electric vehicle capacity mileage data provided in an embodiment of the present application;

[0096] Figure 2 It is a schematic diagram of a reconstructed curve in a power battery capacity prediction method based on electric vehicle capacity mileage data in a specific embodiment of the present application;

[0097] Figure 3 It is a schematic diagram of determining a reference vehicle in a method for predicting power battery capacity based on electric vehicle capacity mileage data in a specific embodiment of the present application;

[0098] Figure 4 This is a functional principle block diagram of a method for predicting the capacity of a power battery based on the capacity mileage data of an electric vehicle to obtain a second capacity in a specific embodiment of the present application;

[0099] Figure 5 It is a schematic diagram of a method for predicting the capacity of a power battery based on the capacity mileage data of an electric vehicle according to a specific embodiment of the present application for obtaining predicted capacity. DETAILED DESCRIPTION

[0100] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0101] like Figure 1 As shown, an embodiment of the present application provides a method for predicting the capacity of a power battery based on the capacity mileage data of an electric vehicle, comprising the following steps:

[0102] Step S100, obtaining the mileage to be predicted of the target vehicle, and determining the historical mileage capacity data of the target vehicle and the overall mileage capacity data of the reference vehicle corresponding to the target vehicle; wherein the mileage to be predicted is greater than the maximum mileage in the historical mileage capacity data, and the mileage to be predicted is less than the maximum mileage in the overall mileage capacity data.

[0103] Step S200: determining a first capacity according to the mileage to be predicted and the overall mileage capacity data.

[0104] Step S300: Determine the second capacity based on the multi-output model according to the historical mileage capacity data, the mileage to be predicted and the overall mileage capacity data.

[0105] Step S400: Determine the predicted capacity corresponding to the mileage to be predicted according to the first capacity and the second capacity.

[0106] Specifically, the target vehicle refers to the vehicle to be predicted, and the mileage to be predicted refers to the mileage to be predicted of the target vehicle. The mileage to be predicted can be the mileage that the target vehicle has reached, or it can be the mileage that the target vehicle has not yet reached. Mileage capacity data refers to the data on the capacity of the power battery under the vehicle mileage. With the use of electric vehicles, the mileage gradually increases, and the capacity of the power battery of electric vehicles gradually decays. The mileage capacity data includes: mileage and the capacity corresponding to the mileage. The historical mileage capacity data refers to the mileage capacity data of the target vehicle in the past. Usually, the historical mileage capacity data counts the mileage capacity data within a certain period of use. The use time corresponding to the mileage to be predicted is later than the use time corresponding to the historical mileage capacity data. For example, the historical mileage capacity data is the mileage capacity data within 20,000 kilometers, and the maximum mileage is 20,000 kilometers. The mileage to be predicted is greater than the maximum mileage, that is, greater than 20,000 kilometers. For example, the mileage to be predicted can be 22,000 kilometers. Then, based on the historical mileage capacity data within the first 20,000 kilometers, the capacity at the mileage to be predicted (such as 22,000 kilometers) is predicted. The reference vehicle refers to the vehicle that provides reference when assisting the prediction of the target vehicle. The overall mileage capacity data refers to the mileage capacity data of the reference vehicle during its entire service life. The service life can be the rated life or the actual life. The rated life refers to the life when the capacity of the vehicle reaches the rated proportion of the initial capacity. The rated proportion can be set as needed, for example, the rated proportion is set to 60%-85%, usually the rated proportion is set to 80%, and the actual life refers to the life when the vehicle is no longer used. The maximum mileage in the overall mileage capacity data can be 600,000-700,000 kilometers.

[0107] The first capacity refers to the capacity predicted based on the overall mileage capacity data of the reference vehicle, and the second capacity refers to the capacity output based on the multi-output model. The multi-output model refers to an output model that outputs two or more data. The first capacity and the second capacity are obtained in two different ways, and then the predicted capacity corresponding to the predicted mileage is determined by the first capacity and the second capacity. The second capacity obtained based on the multi-output model does not require too much overall mileage capacity data of the reference vehicle, and the intrinsic correlation between the aging trajectory of the power battery in the reference vehicle and the target vehicle can be established to predict the capacity of the target vehicle. The first capacity obtained based on the overall mileage capacity data is usually closer to the actual capacity in the actual application scenario of the electric vehicle. The predicted capacity is obtained by combining the first capacity and the second capacity, and the second capacity is corrected by the first capacity, which ensures the reliability of the predicted capacity and is applicable to target vehicles with different user habits and different working conditions, and has strong versatility.

[0108] Step S100 specifically includes:

[0109] Step S110, obtaining vehicle data of the target vehicle and vehicle data of several candidate vehicles; wherein the vehicle data includes several mileage intervals and the capacity corresponding to each mileage in the mileage interval.

[0110] Step S120: Based on the capacity decay model and according to the vehicle data of the target vehicle, determine the historical mileage capacity data of the target vehicle.

[0111] Step S130: Based on the capacity decay model and according to the vehicle data of the candidate vehicle, determine the candidate mileage capacity data of the candidate vehicle.

[0112] Step S140: Based on the historical mileage capacity data, determine the overall mileage capacity data from a number of candidate mileage capacity data, and use the candidate vehicles corresponding to the overall mileage capacity data as reference vehicles corresponding to the target vehicle.

[0113] Specifically, a candidate vehicle refers to a vehicle that is a candidate for auxiliary target vehicle prediction. A candidate vehicle can be any electric vehicle. The more candidates there are and the more models there are, the more conducive it is to determine a reference vehicle that is compatible with the target vehicle from the candidate vehicles, thereby obtaining a more accurate first capacity. Vehicle data includes mileage and capacity. The mileage data is recorded by the vehicle's sensors (such as the odometer), and the capacity is obtained from the vehicle big data platform. If the amount of data obtained on the platform is insufficient, the SOH (State of Health, battery health status) estimation algorithm can be used to fill in the capacity mileage data. SOH = current capacity / initial capacity.

[0114] In order to facilitate the processing of mileage and capacity, the mileage is divided into several mileage intervals. The unit mileage is L mileage, and the number of mileage intervals is l+1. The mileage interval sequence can be expressed as [0, L, 2L, ..., lL]. When the mileage is 0, the capacity of the vehicle is the initial capacity. There are multiple capacities in each mileage interval. When determining the capacity corresponding to the mileage interval, it is necessary to first remove the capacity outliers in the mileage interval. The following steps are used to remove the capacity outliers:

[0115] A100. Determine the first quartile and the third quartile of the volume within the mileage interval.

[0116] A200, determine the outlier range based on the first quartile and the third quartile.

[0117] A300. Based on the abnormal value range, remove the abnormal values ​​in the capacity within the mileage interval, and update the capacity corresponding to each mileage in the mileage interval.

[0118] Specifically, the capacity in the mileage interval is arranged from large to small, and there are three quartiles between the maximum and minimum values ​​of the capacity in the mileage interval. According to the difference between the first quartile Q1 and the third quartile Q3 (Q1>Q3), the range of outliers is determined, that is, (Q1+k(Q1-Q3), +∞) or (-∞, Q3-k(Q1-Q3)), k represents the coefficient, +∞ represents positive infinity, and -∞ represents negative infinity. When the capacity is greater than Q1+k(Q1-Q3) or the capacity is less than Q3-k(Q1-Q3), the capacity is regarded as an outlier.

[0119] After removing the outliers in the capacity, all mileage intervals and the capacity corresponding to each mileage in each mileage interval can be obtained, and then the abnormal vehicles in the candidate vehicles are removed. Abnormal vehicles refer to vehicles with abnormal capacity data. Abnormal capacity data include but are not limited to: abnormal capacity recovery, too little capacity data, uneven distribution of capacity data with mileage, etc. The vehicle data of the target vehicle is obtained after outliers are removed, and used as the vehicle data of the target vehicle. The vehicle data of the candidate vehicle is used as the vehicle data of the candidate vehicle after outliers are removed and / or abnormal vehicles are removed. Since the maximum mileage of the candidate vehicle is greater than the maximum mileage of the target vehicle, the number of mileage intervals of the candidate vehicle is greater than the number of mileage intervals of the target vehicle.

[0120] Based on the capacity decay model, the vehicle data of the target vehicle is processed to obtain the historical mileage capacity data of the target vehicle; based on the capacity decay model, the vehicle data of the candidate vehicle is processed to obtain the candidate mileage capacity data of the candidate vehicle. Then, a reference vehicle is determined from several candidate vehicles. Specifically, it is necessary to compare the historical mileage capacity data with several candidate mileage capacity data. When the capacity decay trajectory of the historical mileage capacity data is similar or close to the capacity decay trajectory of the candidate mileage capacity data, the candidate mileage capacity data is used as the overall mileage capacity data, and the candidate vehicle corresponding to the candidate mileage capacity data is used as the reference vehicle, so as to accurately predict the capacity decay of the target vehicle and obtain an accurate first capacity. Usually, the target vehicle corresponds to one reference vehicle. Of course, the number of reference vehicles corresponding to the target vehicle can also be adjusted as needed.

[0121] The historical mileage capacity data includes: a number of historically reconstructed mileage intervals and the historically reconstructed capacities corresponding to the historically reconstructed mileage intervals. Step S120 specifically includes:

[0122] Step S121: Based on the capacity decay model, for each mileage interval of the target vehicle, the mileage within the mileage interval and the corresponding capacity are fitted to obtain a target reconstruction curve.

[0123] Step S122: Determine the historical reconstruction mileage interval and the historical reconstruction capacity corresponding to the historical reconstruction mileage interval based on the target reconstruction curve.

[0124] Specifically, the historical mileage capacity data of the target vehicle can be divided into several mileage intervals, the unit mileage is N mileage, and the number of mileage intervals is n+1. Then the mileage interval sequence can be expressed as [0, N, 2N, ..., nN]. When the mileage is 0, the capacity of the vehicle is the initial capacity. The capacity in each mileage interval is fitted by a piecewise fitting method. After all mileage intervals are fitted, the target reconstruction curve is obtained. After the target reconstruction curve is obtained, several historical reconstruction mileage intervals and corresponding historical reconstruction capacities are determined according to the target reconstruction curve. The number of historical reconstruction mileage intervals and the number of mileage intervals (n+1) can be the same or different. For example, if the number of historical reconstruction mileage intervals is the same as the number of mileage intervals (n+1), the historical reconstruction mileage interval sequence [0, N, 2N, ..., nN] and the historical reconstruction capacity sequence [y(0), y(N), y(2N), ..., y(nN)] can be obtained, where y(·) represents the capacity.

[0125] The candidate mileage capacity data includes: a number of candidate reconstructed mileage intervals and candidate reconstructed capacities corresponding to the candidate reconstructed mileage intervals. Step S130 specifically includes:

[0126] Step S131: Based on the capacity decay model, for each mileage interval of the candidate vehicle, the mileage within the mileage interval and the corresponding capacity are fitted to obtain a candidate reconstruction curve.

[0127] Step S132: Determine the candidate reconstruction mileage interval and the candidate reconstruction capacity corresponding to the candidate reconstruction mileage interval based on the candidate reconstruction curve.

[0128] Specifically, the historical mileage capacity data of the candidate vehicle can be divided into several mileage intervals, with a unit mileage of N mileage, and the number of mileage intervals is m+1. Then the mileage interval sequence can be expressed as [0, N, 2N, ..., mN]. When the mileage is 0, the capacity of the vehicle is the initial capacity, and l and m can be equal or unequal. The capacity in each mileage interval is fitted by a piecewise fitting method. After all mileage intervals are fitted, a candidate reconstruction curve is obtained. After obtaining the candidate reconstruction curve, several candidate reconstruction mileage intervals and corresponding candidate reconstruction capacities are determined according to the candidate reconstruction curve. The number of candidate reconstruction mileage intervals and the number of mileage intervals (m+1) can be the same or different. For example, if the number of candidate reconstruction mileage intervals is the same as the number of mileage intervals (m+1), then the candidate reconstruction mileage interval sequence [0, N, 2N, ..., mN] and the candidate reconstruction capacity sequence [y(0), y(N), y(2N), ..., y(mN)] can be obtained, n <m。

[0129] like Figure 2 As shown, the corresponding reconstructed curve is obtained based on the vehicle data, and the data of the reconstructed curve is easy to process. The target reconstructed curve of the target vehicle can be obtained according to the vehicle data of the target vehicle, and the candidate reconstructed curve can be obtained according to the vehicle data of the candidate vehicle. In order to facilitate data processing, the vehicle data is normalized, for example, a calibration normalization method is adopted, a calibration capacity is configured, and the quotient of the capacity and the calibration capacity is used as the normalized capacity. The reconstructed curve is obtained by fitting according to the mileage and the corresponding normalized capacity.

[0130] The capacity fading model is:

[0131] y=ae bx +cx 2 +d;

[0132] Among them, y represents capacity, e represents natural constant, a, b, c, d represent model parameters, x represents mileage, and the maximum mileage in the mileage range is taken as the mileage.

[0133] Specifically, the capacity decay model uses exponential and quadratic polynomials to fit the mileage and capacity. Although different mileage intervals are based on the same capacity decay model, the model parameters obtained by fitting are different, which makes the fitting accuracy better. In the fitting process, an optimization algorithm can be used to achieve the optimization task. The optimization problem can be expressed as:

[0134]

[0135] Here, argmin(·) represents the variable value when the objective function takes the minimum value.

[0136] The objective function is a residual function, and a suitable optimization algorithm is used to minimize the residual function. The optimization algorithm can be a nonlinear optimization algorithm, for example, a Levenberg-Marquardt algorithm, a quasi-Newton method (such as a BFGS algorithm), etc. The optimization algorithm updates the parameter estimates through iteration so that the residual function approaches the minimum value. The initial values ​​of the model parameters can be determined based on experience or preliminary analysis. The selected optimization algorithm inputs the initial values ​​of the model parameters and iteratively updates the model parameters until the convergence condition is reached. The convergence condition can be that the residual function value is less than a preset threshold or the number of iterations reaches the maximum number of iterations.

[0137] The overall mileage capacity data includes: a number of overall reconstructed mileage intervals and the overall reconstructed capacity corresponding to the overall reconstructed mileage intervals. Step S140 specifically includes:

[0138] Step S141: for each candidate vehicle, determine the cumulative distance cost of the candidate vehicle; wherein the cumulative distance cost is the cumulative value of the distances between a number of historically reconstructed mileage intervals and their corresponding historically reconstructed capacities, and a number of candidate reconstructed mileage intervals and their corresponding candidate reconstructed capacities.

[0139] Step S142: taking the candidate reconstructed mileage interval corresponding to the minimum value of the cumulative distance cost of the candidate vehicles as the overall reconstructed mileage interval, and taking the candidate reconstructed capacity corresponding to the minimum value of the cumulative distance cost of the candidate vehicles as the overall reconstructed capacity.

[0140] Specifically, the DTW (Dynamic Time Warping) distance is used as a basis to determine the reference vehicle corresponding to the target vehicle from a number of candidate vehicles. Of course, other measurement criteria can also be used as a basis to determine the reference vehicle corresponding to the target vehicle from a number of candidate vehicles. For each candidate vehicle, the cumulative distance cost of the candidate vehicle is calculated. The smaller the cumulative distance cost, the smaller the DTW distance between the historical reconstructed mileage interval sequence and the candidate reconstructed capacity sequence, and the closer the similarity between the historical reconstructed mileage interval sequence and the candidate reconstructed capacity sequence. The candidate vehicle corresponding to the minimum value of the cumulative distance cost can be used as the reference vehicle. Figure 3 As shown, the black vehicle represents the target vehicle, and the target vehicle has a target reconstruction curve (solid line). The gray vehicle represents the candidate vehicle, and the candidate vehicle has a candidate reconstruction curve (dashed line). The candidate vehicle with the smallest DTW distance is used as the reference vehicle of the target vehicle.

[0141] The cumulative distance cost is:

[0142] C(i,j)=D(i,j)+min(C(i-1,j),C(i,j-1),C(i-1,j-1));

[0143] C(0,0) <C(0,j)=C(i,0)=∞;

[0144] Among them, C(i,j) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-th candidate reconstruction mileage interval, D(i,j) represents the distance between the historical reconstruction capacity corresponding to the i-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-th candidate reconstruction mileage interval, C(i-1,j) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-1-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-1-th candidate reconstruction mileage interval, C(i,j-1) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-1-th candidate reconstruction mileage interval, C(i-1,j-1) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-1-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-1-th candidate reconstruction mileage interval, and min(·) represents the minimum value operation.

[0145] Specifically, D(i,j) represents the distance between the historical reconstruction capacity y(iN) corresponding to the i-th historical reconstruction mileage interval iN and the candidate reconstruction capacity y(iN) corresponding to the j-th candidate reconstruction mileage interval jN. The distance can be calculated using the Euclidean distance. The cumulative distance costs C(n,m) calculated for different candidate vehicles are different. The candidate vehicle corresponding to the minimum value of all cumulative distance costs C(n,m) is used as the reference vehicle.

[0146] Step S200 specifically includes:

[0147] Step S210: Determine the overall reconstructed mileage interval corresponding to the mileage to be predicted according to the mileage to be predicted and a plurality of overall reconstructed mileage intervals.

[0148] Step S220: taking the overall reconstructed capacity corresponding to the overall reconstructed mileage interval as the first capacity; or, determining the capacity corresponding to the mileage to be predicted according to the model parameters of the overall reconstructed mileage interval in the fitting process and the mileage to be predicted, and taking the capacity as the first capacity.

[0149] Specifically, after determining the reference vehicle and the overall mileage capacity data of the reference vehicle, the corresponding overall reconstructed mileage interval or the overall reconstructed mileage within the overall reconstructed mileage interval can be determined according to the mileage to be predicted, and the overall reconstructed capacity corresponding to the overall reconstructed mileage interval can be used as the first capacity, or the overall reconstructed capacity corresponding to the overall reconstructed mileage can be used as the first capacity. Of course, it is also possible to calculate the capacity corresponding to the mileage to be predicted based on the model parameters of the mileage to be predicted and the overall reconstructed mileage interval during the fitting process (or the candidate reconstruction curve corresponding to the overall reconstructed mileage interval), and use this capacity as the first capacity. Here, the expression of the candidate reconstruction curve is y=ae bx +cx 2 +d, the predicted mileage is input as x, and the obtained capacity y is used as the first capacity.

[0150] Step S300 specifically includes:

[0151] Step S310: Train the multi-output model according to the historical reconstructed mileage interval, the historical reconstructed capacity, the overall reconstructed mileage interval and the overall reconstructed capacity to obtain a trained multi-output model.

[0152] Step S320: input the mileage to be predicted into the trained multi-output model, and output the second capacity through the trained multi-output model.

[0153] Specifically, Figure 4 As shown, based on the historical reconstructed mileage interval and historical reconstructed capacity of the target vehicle, the overall reconstructed mileage interval and overall reconstructed capacity of the reference vehicle, the multi-output model is trained to obtain a trained multi-output model, and the mileage to be predicted is input into the trained multi-output model to output the second capacity. The multi-output model is used to output the second capacity, which reflects the capacity data with a high probability for most vehicles, which is convenient for predicting application scenarios with different user habits and different working conditions, and improves the versatility of the prediction method.

[0154] Specifically, the multi-output model can adopt multi-output Gaussian process regression, and the multi-output model is:

[0155] y~GP(m(x),K M );

[0156]

[0157] K x =(k pq (xx′));

[0158]

[0159] Where y represents capacity, which can be historical reconstruction capacity or overall reconstruction capacity, GP(·,·) represents Gaussian process, m(·) represents mean function, and K M represents the covariance function, K f Represents a relationship matrix, which represents the correlation between the output capacities. The elements on the diagonal of the relationship matrix represent the correlation between the output capacity and itself, and the elements on the off-diagonal represent the correlation between the capacities of different outputs. The initial value of the relationship matrix can be the unit matrix, indicating that there is no correlation between the capacities of the outputs. represents the Kronecker product, K x Represents the correlation matrix of mileage. The correlation matrix of mileage represents the correlation between the input mileages and is calculated using the Matern function. k pq (xx′) represents K x elements in , σ f , ρ are hyperparameters, υ is a smoothing parameter, and the value of the smoothing parameter is configured as needed. For example, the smoothing parameter v = 5 / 2, R υ (·) represents the corrected Bessel function, x and x′ both represent the mileage, the maximum mileage in the historical reconstructed mileage interval or the maximum mileage in the overall reconstructed mileage interval is taken as the mileage, and Γ(·) represents the gamma function.

[0160] Specifically, K f is a t*t order matrix, t represents the number of vehicles, K x It is a w*w matrix, where w represents the length of the mileage sequence. It is also possible to establish a negative log-likelihood function of the conditional probability of the training sample, find the partial derivative of the hyperparameters, and then use optimization methods such as the conjugate gradient method or Newton's method to minimize the partial derivatives to obtain the optimal solution for the hyperparameters, thereby obtaining the posterior distribution of the second capacity.

[0161] The predicted capacity is:

[0162] y pre =α*y1+(1-α)*y2;

[0163] Among them, y pre represents the predicted capacity, α represents the weight, α<1, y1 represents the first capacity, and y2 represents the second capacity.

[0164] Specifically, the first capacity and the second capacity are weighted and summed to obtain the predicted capacity, and the weight can be configured as needed. Combining the first capacity and the second capacity can make the predicted capacity more reliable. Figure 5 As shown, the predicted capacity obtained by the prediction method of the present invention is shown by the dotted line.

[0165] The historical mileage capacity data of the target vehicle includes the initial capacity; the power battery capacity prediction method also includes:

[0166] Step S400: Determine the remaining life of the power battery of the target vehicle according to the initial capacity and the predicted capacity.

[0167] Specifically, the remaining life of the power battery refers to the remaining mileage when the capacity of the power battery reaches the rated proportion of the initial capacity. The remaining life of the power battery of the target vehicle can be determined based on the initial capacity and the predicted capacity. When the capacity of the vehicle reaches the rated proportion of the initial capacity, it indicates that the remaining mileage is 0 and the remaining life of the power battery is 0; when the capacity of the vehicle does not reach the rated proportion of the initial capacity and a certain mileage Δy is required for the capacity of the vehicle to reach the rated proportion of the initial capacity, it indicates that the remaining mileage is Δy and the remaining life of the power battery is Δy. Specifically, if the predicted capacity is the rated proportion of the initial capacity, the remaining mileage is obtained by subtracting the maximum mileage in the historical mileage capacity data from the mileage to be predicted corresponding to the predicted capacity, and the remaining mileage can be used as the remaining life of the power battery of the target vehicle. The rated proportion is set to 60%-85%, and the rated proportion is usually set to 80%.

[0168] Steps S100 to S300 and step S400 may be performed entirely by the vehicle, or entirely by the server or terminal, or partially by the vehicle and partially by the server or terminal. The present invention is not limited in terms of the execution subject, as long as the actions disclosed in the embodiments of the present invention are performed. The server includes an independent physical server, a physical server cluster or a virtual server, for example, the server may be a cloud server. The terminal includes a desktop terminal or a mobile terminal, such as a desktop computer, a tablet computer, a laptop computer, a smart phone, etc.

[0169] The advantages of the present invention are as follows:

[0170] 1. The present invention uses the mileage corresponding to the capacity as the basis for determining the capacity of the electric vehicle power battery, rather than determining the capacity or remaining life of the power battery based on time. The present invention is more practical, and the outputted remaining life is more valuable for reference to the driver.

[0171] 2. The present invention can predict the capacity or remaining life of the power battery of the target vehicle by only using the historical mileage capacity data of the target vehicle and the overall mileage capacity data of the reference vehicle, without the need for tedious preprocessing and feature engineering of complex, fragmented and irregular real vehicle data.

[0172] 3. The present invention reconstructs the mileage capacity data of the electric vehicle, and extracts a reconstructed curve that can represent the decreasing trend of the electric vehicle power battery capacity from the scattered original data, thereby avoiding the problem that the mileage capacity data cannot be directly used due to the quality of the original data.

[0173] 4. The present invention can automatically match the capacity attenuation trajectory of a reference vehicle that has a similar power battery aging pattern to the target vehicle, and use the inherent correlation between the aging trajectories of the reference vehicle and the target vehicle to predict the future life of the target vehicle. It is applicable to automobile data generated by various user habits and different working conditions, and has strong versatility.

[0174] 5. The present invention uses a fusion of multiple algorithms to predict the capacity or remaining life of the electric vehicle power battery, thereby ensuring the reliability of the prediction results.

[0175] Based on the power battery capacity prediction method based on electric vehicle capacity mileage data of any of the above embodiments, an embodiment of the present application further provides a power battery capacity prediction device based on electric vehicle capacity mileage data, comprising:

[0176] An acquisition module is used to acquire the mileage to be predicted of the target vehicle, and determine the historical mileage capacity data of the target vehicle and the overall mileage capacity data of the reference vehicle corresponding to the target vehicle; wherein the time corresponding to the mileage to be predicted is later than the time corresponding to the historical mileage capacity data, and the mileage to be predicted is less than the maximum mileage in the overall mileage capacity data;

[0177] A first capacity module, used for determining a first capacity according to the mileage to be predicted and the overall mileage capacity data;

[0178] A second capacity module, for determining a second capacity based on a multi-output model according to historical mileage capacity data, mileage to be predicted, and overall mileage capacity data;

[0179] The prediction module is used to determine the predicted capacity corresponding to the mileage to be predicted according to the first capacity and the second capacity.

[0180] Based on the power battery capacity prediction method based on electric vehicle capacity mileage data in any of the above embodiments, an embodiment of the present application further provides a vehicle-mounted terminal. The vehicle-mounted terminal may include:

[0181] A memory, a processor, and a computer program stored in the memory and executable on the processor.

[0182] When the processor executes the program, the power battery capacity prediction method based on the electric vehicle capacity mileage data provided in the above embodiment is implemented.

[0183] Furthermore, the vehicle-mounted terminal also includes:

[0184] Communication interface, used for communication between memory and processor.

[0185] The memory may include a high-speed RAM (Random Access Memory) memory, and may also include a non-volatile memory, such as at least one disk memory.

[0186] If the memory, processor and communication interface are implemented independently, the communication interface, memory and processor can be connected to each other through a bus and communicate with each other. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0187] Specifically, if the memory, processor and communication interface are integrated on a chip, the memory, processor and communication interface can communicate with each other through an internal interface.

[0188] The processor may be a CPU (Central Processing Unit), or an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present application.

[0189] In addition, based on any one of the above-mentioned embodiments of the power battery capacity prediction device based on the electric vehicle capacity mileage data or the vehicle-mounted terminal, an embodiment of the present application also proposes a vehicle, which includes the power battery capacity prediction device based on the electric vehicle capacity mileage data of the above-mentioned embodiment, or the vehicle-mounted terminal of the above-mentioned embodiment.

[0190] In the description of this specification, the description with reference to the terms "embodiment", "any embodiment" or "implementation" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or implementation are included in at least one embodiment or implementation of the present application. In this specification, the schematic representations of the above terms are not necessarily directed to the same embodiment or implementation. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or implementations in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or implementations described in this specification and the features of the different embodiments or implementations, unless they are contradictory.

[0191] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features.

[0192] It should be understood that the various parts of the present application can be implemented in hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array, a field programmable gate array, etc.

[0193] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

Claims

1. A method for predicting power battery capacity based on electric vehicle capacity mileage data, characterized in that: Includes steps: Acquire the mileage to be predicted of the target vehicle, and determine the historical mileage capacity data of the target vehicle and the overall mileage capacity data of the reference vehicle corresponding to the target vehicle; wherein the mileage to be predicted is greater than the maximum mileage in the historical mileage capacity data, and the mileage to be predicted is less than the maximum mileage in the overall mileage capacity data; Determining a first capacity according to the mileage to be predicted and the overall mileage capacity data; Determining a second capacity based on the multi-output model according to the historical mileage capacity data, the mileage to be predicted, and the overall mileage capacity data; A predicted capacity corresponding to the mileage to be predicted is determined according to the first capacity and the second capacity.

2. The method for predicting power battery capacity based on electric vehicle capacity mileage data according to claim 1, characterized in that: The determining of the historical mileage capacity data of the target vehicle and the overall mileage capacity data of the reference vehicle corresponding to the target vehicle comprises: Acquire vehicle data of a target vehicle and vehicle data of several candidate vehicles; wherein the vehicle data includes several mileage intervals and the capacity corresponding to each mileage in the mileage intervals; Based on the capacity decay model, according to the vehicle data of the target vehicle, determine the historical mileage capacity data of the target vehicle; Determining candidate mileage capacity data of the candidate vehicle based on the capacity decay model and according to the vehicle data of the candidate vehicle; Based on the historical mileage capacity data, the overall mileage capacity data is determined from a plurality of candidate mileage capacity data, and the candidate vehicles corresponding to the overall mileage capacity data are used as reference vehicles corresponding to the target vehicle.

3. The method for predicting the power battery capacity based on the electric vehicle capacity mileage data according to claim 2 is characterized in that: The historical mileage capacity data includes: a plurality of historically reconstructed mileage intervals and historically reconstructed capacities corresponding to the historically reconstructed mileage intervals; The method of determining the historical mileage capacity data of the target vehicle based on the capacity decay model and according to the vehicle data of the target vehicle includes: Based on the capacity decay model, for each mileage interval of the target vehicle, the mileage and the corresponding capacity in the mileage interval are fitted to obtain a target reconstruction curve; Based on the target reconstruction curve, a historical reconstruction mileage interval and a historical reconstruction capacity corresponding to the historical reconstruction mileage interval are determined.

4. The method for predicting power battery capacity based on electric vehicle capacity mileage data according to claim 3, characterized in that: The candidate mileage capacity data includes: a plurality of candidate reconstruction mileage intervals and candidate reconstruction capacities corresponding to the candidate reconstruction mileage intervals; The determining, based on the capacity decay model and according to the vehicle data of the candidate vehicle, the candidate mileage capacity data of the candidate vehicle comprises: Based on the capacity decay model, for each mileage interval of the candidate vehicle, the mileage in the mileage interval and the corresponding capacity are fitted to obtain a candidate reconstruction curve; Based on the candidate reconstruction curve, a candidate reconstruction mileage interval and a candidate reconstruction capacity corresponding to the candidate reconstruction mileage interval are determined.

5. The method for predicting the power battery capacity based on the electric vehicle capacity mileage data according to claim 4 is characterized in that: The capacity decay model is: y=ae bx +cx 2 +d; Among them, y represents capacity, e represents natural constant, a, b, c, d represent model parameters, and x represents mileage.

6. The method for predicting power battery capacity based on electric vehicle capacity mileage data according to claim 5, characterized in that: The overall mileage capacity data includes: a plurality of overall reconstructed mileage intervals and overall reconstructed capacities corresponding to the overall reconstructed mileage intervals; The determining of the overall mileage capacity data from a plurality of candidate mileage capacity data based on the historical mileage capacity data comprises: For each candidate vehicle, determine the cumulative distance cost of the candidate vehicle; wherein the cumulative distance cost is the cumulative value of the distances between a number of the historical reconstructed mileage intervals and their corresponding historical reconstructed capacities, and a number of the candidate reconstructed mileage intervals and their corresponding candidate reconstructed capacities; The candidate reconstructed mileage interval corresponding to the minimum value of the cumulative distance cost of the candidate vehicles is used as the overall reconstructed mileage interval, and the candidate reconstructed capacity corresponding to the minimum value of the cumulative distance cost of the candidate vehicles is used as the overall reconstructed capacity.

7. The method for predicting the power battery capacity based on the electric vehicle capacity mileage data according to claim 6, characterized in that: The cumulative distance cost is: C(i,j)=D(i,j)+min(C(i-1,j), C(i,j-1), C(i-1,j-1)); C(0,0) <C(0,j)=C(i,0)=∞; Among them, C(i, j) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-th candidate reconstruction mileage interval, D(i, j) represents the distance between the historical reconstruction capacity corresponding to the i-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-th candidate reconstruction mileage interval, C(i-1, j) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-1-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-1-th candidate reconstruction mileage interval, C(i, j-1) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-1-th candidate reconstruction mileage interval, C(i-1, j-1) represents the cumulative distance cost between the historical reconstruction capacity corresponding to the i-1-th historical reconstruction mileage interval and the candidate reconstruction capacity corresponding to the j-1-th candidate reconstruction mileage interval, and min(·) represents the minimum value operation.

8. The method for predicting power battery capacity based on electric vehicle capacity mileage data according to claim 6, characterized in that: The determining the first capacity according to the mileage to be predicted and the overall mileage capacity data includes: Determining the overall reconstructed mileage interval corresponding to the mileage to be predicted according to the mileage to be predicted and a plurality of the overall reconstructed mileage intervals; The overall reconstructed capacity corresponding to the overall reconstructed mileage interval is used as the first capacity; or, according to the model parameters of the overall reconstructed mileage interval in the fitting process and the mileage to be predicted, the capacity corresponding to the mileage to be predicted is determined, and the capacity is used as the first capacity.

9. The method for predicting the power battery capacity based on the electric vehicle capacity mileage data according to any one of claims 6 to 8, characterized in that: The determining the second capacity based on the multi-output model according to the historical mileage capacity data, the mileage to be predicted and the overall mileage capacity data includes: Training a multi-output model according to the historically reconstructed mileage interval, the historically reconstructed capacity, the overall reconstructed mileage interval, and the overall reconstructed capacity to obtain a trained multi-output model; The mileage to be predicted is input into the trained multi-output model, and the second capacity is output through the trained multi-output model.

10. The method for predicting power battery capacity based on electric vehicle capacity mileage data according to claim 9, characterized in that: The multi-output model is: y~GP(m(x),K M ); K x =(k pq (x-x′)); Where y represents capacity, GP(·,·) represents Gaussian process, m(·) represents mean function, and K M represents the covariance function, K f represents the relationship matrix, represents the Kronecker product, K x represents the correlation matrix of mileage, k pq (xx′) represents K x Elements in , σ f , ρ are hyperparameters, υ is a smoothing parameter, R υ (·) represents the modified Bessel function, x and x′ represent the mileage, and Γ(·) represents the gamma function.

11. The method for predicting power battery capacity based on electric vehicle capacity mileage data according to any one of claims 1 to 8, characterized in that: The predicted capacity is: y pre =α*y1+(1-α)*y2; Among them, y pre represents the predicted capacity, α represents the weight, y1 represents the first capacity, and y2 represents the second capacity.

12. The method for predicting the capacity of a power battery based on the capacity mileage data of an electric vehicle according to any one of claims 1 to 8, characterized in that: The historical mileage capacity data of the target vehicle includes an initial capacity; The power battery capacity prediction method further comprises the steps of: The remaining life of the power battery of the target vehicle is determined according to the initial capacity and the predicted capacity.

13. A power battery capacity prediction device based on electric vehicle capacity mileage data, characterized in that: include: an acquisition module, used for acquiring the mileage to be predicted of the target vehicle, and determining the historical mileage capacity data of the target vehicle and the overall mileage capacity data of the reference vehicle corresponding to the target vehicle; wherein the time corresponding to the mileage to be predicted is later than the time corresponding to the historical mileage capacity data, and the mileage to be predicted is less than the maximum mileage in the overall mileage capacity data; A first capacity module, configured to determine a first capacity according to the mileage to be predicted and the overall mileage capacity data; A second capacity module, configured to determine a second capacity based on a multi-output model according to the historical mileage capacity data, the mileage to be predicted and the overall mileage capacity data; A prediction module is used to determine the predicted capacity corresponding to the mileage to be predicted according to the first capacity and the second capacity.

14. A vehicle-mounted terminal, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for predicting the capacity of a power battery based on the capacity mileage data of an electric vehicle as described in any one of claims 1 to 12.

15. A vehicle, characterized in that: include: A power battery capacity prediction device based on electric vehicle capacity mileage data as described in claim 13, or a vehicle-mounted terminal as described in claim 14.

Images