Electric vehicle charging strategy identification method, charging duration prediction method and system

Abstract

The application belongs to the technical field of new energy vehicles. A charging strategy identification method, a charging duration prediction method and a system for an electric vehicle are provided. The charging strategy can be quickly identified from long-span historical data in the case of unknown original charging strategy, thereby providing accurate data support for subsequent charging strategy analysis and charging duration prediction. Through multi-stage efficient charging duration prediction of multi-source information fusion, the prediction accuracy of the charging duration is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy vehicles, and in particular to a method for identifying a charging strategy for an electric vehicle, and a method and system for predicting charging duration. Background Art

[0002] The statements in this section merely provide background art related to the present invention and do not necessarily constitute prior art.

[0003] To better plan and manage charging infrastructure, charging station operators need to accurately forecast charging demand across different time periods and locations. Charging time predictions help optimize the configuration and scheduling of charging stations, improving charging station utilization efficiency. Accurate charging time predictions can help grid operators better manage power loads, especially during peak hours, avoiding grid overload and unnecessary power waste. They can also promote the use of renewable energy and balance power supply and demand by optimizing charging schedules. For electric vehicle users, accurate charging time predictions can help them better plan their trips, reduce wait times, and enhance their overall driving experience. Charging time predictions can help develop more optimized charging strategies to extend battery life and reduce charging costs.

[0004] Due to the complex electrochemical reactions inside the battery, the complex external environment, and the changeable operating conditions, it is very difficult to accurately predict the battery charging time. Currently, the charging time prediction methods for power batteries mainly include model methods and deep learning methods.

[0005] The prediction method based on the battery model establishes a physical and chemical model of the battery, simulates the behavior of the battery under different charging conditions, and predicts the charging time. It has a high theoretical basis and can explain the physical and chemical phenomena in the charging process. However, the model establishment and parameter calibration are complex, and the accuracy of the model depends on the understanding of the internal mechanism of the battery and the accuracy of the data.

[0006] Deep learning methods use deep learning models (such as LSTM and GRU) for prediction. They can handle complex nonlinear relationships, have strong generalization capabilities, and can learn underlying patterns from large amounts of data. However, they require a large amount of training data, and the model training and parameter adjustment process is relatively complex, requiring high computing resources, making them difficult to apply industrially.

[0007] Furthermore, existing technologies often rely on known charging strategy details for charge duration prediction, making them inapplicable to unknown charging strategies. For example, the dataset used in this paper lacks full charge and discharge cycles, contains multiple charging strategies, and the original charging strategy (including the charging current switching point) is unknown. Summary of the Invention

[0008] In order to solve the problems in the prior art, the application provides an electric vehicle charging strategy identification method, a charging duration prediction method and a system, which can quickly identify a charging strategy from long-span historical data in the case of unknown original charging strategy, thereby providing accurate data support for subsequent charging strategy analysis and charging duration prediction, and greatly improving the prediction accuracy of charging duration through multi-source information fusion and multi-stage efficient charging duration prediction.

[0009] In order to achieve the above object, the application adopts the following technical scheme:

[0010] In a first aspect, the application provides an electric vehicle charging strategy identification method.

[0011] An electric vehicle charging strategy identification method comprises the following processes:

[0012] Charging segment division is performed according to the cleaned real vehicle data of the electric vehicle, and a feature matrix is constructed according to the charging segment division result and the cleaned real vehicle data of the electric vehicle.

[0013] Current features in the feature matrix are extracted, and first-level clustering of current gears is performed.

[0014] Time features in the feature matrix are extracted, the duration of each charging segment at different gears is counted, and a new feature matrix for representing the duration of each charging segment at different gears is constructed.

[0015] Second-level clustering is performed on the new feature matrix, and then the identification result of typical charging strategies and special charging strategies is obtained.

[0016] As a further limitation of the first aspect of the application, the charging segment division according to the cleaned real vehicle data of the electric vehicle comprises:

[0017] The cleaned real vehicle data is filtered, and the data in the charging state of charging or charging completion is filtered, the filtered data is traversed, for two adjacent data, it is judged whether the absolute value of the SOC difference is greater than 1 and the time difference is greater than five sampling intervals, if yes, the two adjacent data have the same charging segment ID, otherwise, the charging segment ID is increased by 1, the start and end SOC of each charging segment is checked, the charging segment with a start and end SOC difference less than a set threshold is removed, and the final charging segment division result is obtained.

[0018] As a further limitation of the first aspect of the application, the first-level clustering of current gears comprises:

[0019] A data point is randomly selected from the data set, if the data point has been visited, the next data point is selected;

[0020] Count the number of neighbors within the radius of ε of the selected data point, if the data point meets the condition of core point, then form a new cluster with the data point as the core: N ε (P)>MinPts, wherein, ε is the maximum distance of a point within the radius of ε to define a neighbor, MinPts is the minimum number of neighbors required for a point to be a core point;

[0021] Expand all directly density reachable points of the core point, mark them as belonging to the same cluster; the directly density reachable is: if point q is within the ε neighborhood of point p, and point p is a core point, then point q is a directly density reachable point of point p;

[0022] For points that cannot be expanded, mark them as boundary points, and for points that are neither core points nor within the ε neighborhood of any core point, mark them as noise points;

[0023] Calculate the cluster center of each cluster, wherein, C is the center of the cluster, n is the number of points in the cluster, I i is the i-th current value in the cluster.

[0024] As a further limitation of the first aspect of the application, the new feature matrix is subjected to secondary clustering, comprising:

[0025] Randomly select K initial centroids, where the value of K is 2, to divide out the typical charging strategy cluster and the special charging strategy cluster;

[0026] For each charging segment j, calculate its Euclidean distance to each centroid, and assign it to the cluster corresponding to the nearest centroid: wherein, c k is the k-th centroid, c ik is the value of the k-th centroid in the i-th dimension;

[0027] Recalculate the centroid of each cluster: wherein, C k is the set of charging segments in the k-th cluster;

[0028] Repeat the steps of assigning clusters and updating centroids until the centroids no longer change or the maximum number of iterations is reached;

[0029] Each charging segment j will be assigned a cluster label k, indicating the cluster to which the charging segment j belongs.

[0030] As a further limitation of the first aspect of the application, the total current feature in the feature matrix is extracted, the total current feature is subjected to least squares linear fitting, the slope feature and the mutation time are extracted, the data of the minimum monomer temperature is obtained according to the mutation time point, and the probability distribution is calculated;

[0031] The typical charging strategy is an approximate two-stage constant current charging, a first stage is a small current ramp charging and preheating, and when the minimum single cell temperature reaches a set temperature threshold, a second stage constant current charging stage is entered until the power is fully charged.

[0032] In a second aspect, the present application provides an electric vehicle charging strategy identification system.

[0033] An electric vehicle charging strategy identification system comprises:

[0034] A feature matrix construction unit is configured to divide charging segments according to real vehicle data of the cleaned electric vehicle, and construct a feature matrix according to the charging segment division result and the real vehicle data of the cleaned electric vehicle;

[0035] A primary clustering unit is configured to extract current features in the feature matrix, and perform primary clustering of current gears;

[0036] A new feature matrix construction unit is configured to extract time features in the feature matrix, count the duration of each charging segment at different gears, and construct a new feature matrix for representing the duration of each charging segment at different gears;

[0037] A secondary clustering unit is configured to perform secondary clustering on the new feature matrix, and further obtain an identification result of a typical charging strategy and a special charging strategy.

[0038] In a third aspect, the present application provides an electric vehicle charging duration prediction method.

[0039] An electric vehicle charging duration prediction method is used to predict the duration of the typical charging strategy obtained by the electric vehicle charging strategy identification method of the first aspect of the present application, comprising the following processes:

[0040] A robust least squares method is used to establish a temperature increment-temperature increment time relationship to obtain a preheating stage charging duration prediction result;

[0041] A support vector regression method is used to establish a charging SOC-current value relationship, and combined with the battery capacity, a charging duration prediction result of the second-stage constant current charging stage is obtained;

[0042] After the end of the second-stage constant current charging stage, when the SOC reaches 99%, a small current is used for energy supplement to obtain a supplement time;

[0043] The sum of the preheating stage charging duration prediction result, the charging duration prediction result of the second-stage constant current charging stage and the supplement time is taken as the final charging duration prediction result.

[0044] As part of the third aspect is further limited, the temperature increment-temperature increment time relationship is established by using robust least squares method, and the preheating stage charging time prediction result is obtained, including: t1=a*ΔT+b, ΔT=T end -T0, ΔT is the temperature increment, T0 and T end are the minimum temperature of the single body at the start of the preheating stage, t1 is the preheating stage time, and a and b are parameters to be fitted.

[0045] The support vector regression method is used to establish the relationship between the charging SOC and the current value: I0=ω*SOC0+b, wherein I0 is the second stage starting current, SOC0 is the second stage starting SOC, ω is the weight vector, and b is the bias term.

[0046] The charging time prediction result of the second constant current charging stage is: Wherein, ΔSOC is the start and end SOC difference; I1 is the end current, which is a constant; Q fin is the battery capacity.

[0047] The energy supplement time is Wherein, I end is the energy supplement current.

[0048] As part of the third aspect is further limited, the battery capacity Q fin is obtained, including:

[0049] The longest current stable charging segment is selected, and the current charging segment battery capacity is calculated by using the inverse ampere-hour integral method: Wherein, Q is the battery capacity, I(t) is the current at time t, SOC(t0) and SOC(t1) are the start and end SOC of the charging segment, t0 and t1 are the start and end time of the charging segment.

[0050] A continuous charging process is randomly intercepted from the current charging segment, intercepted n times, and the charging segment capacity is calculated respectively, and the mean value is selected as the current charging segment capacity: Wherein, Q i is the battery capacity calculated each time, and Q fin is the final battery capacity.

[0051] The Savitzky-Golay filter is added, the robust least squares method is used to fit the capacity-time curve, and Q fin =a*Time+b, wherein Time is the time, and a and b are parameters to be fitted.

[0052] In the fourth aspect, the application provides a charging time prediction system for an electric vehicle.

[0053] A kind of electric vehicle charging duration prediction system, the typical charging strategy identified to the electric vehicle charging strategy identification method described in the first aspect of the present application carries out duration prediction, comprising:

[0054] First prediction unit is configured to: adopt robust least square method to establish temperature increment-temperature increment time relationship, obtains the preheating stage charging duration prediction result;

[0055] Second prediction unit is configured to: adopt support vector regression method to establish the relationship between charging SOC and current value, combined with battery capacity, obtains the charging duration prediction result of two-stage constant current charging stage;

[0056] Supplement energy time acquisition unit is configured to: after two-stage constant current charging stage ends, when SOC reaches 99%, small current is used to supplement energy, and supplement energy time is obtained;

[0057] Comprehensive prediction unit is configured to: the sum of the preheating stage charging duration prediction result, the charging duration prediction result of two-stage constant current charging stage and supplement energy time is used as the final charging duration prediction result.

[0058] Compared with prior art, the beneficial effects of the present application are:

[0059] 1、The present application innovatively proposes an electric vehicle charging strategy identification method, which can quickly identify the charging strategy from long-span historical data under unknown original charging strategy, thereby providing accurate data support for subsequent charging strategy analysis and charging duration prediction.

[0060] 2、The present application innovatively proposes an electric vehicle charging duration prediction method, which greatly improves the prediction accuracy of charging duration through multi-source information fusion multi-stage efficient charging duration prediction.

[0061] 3、The real capacity of the whole vehicle battery is difficult to measure, therefore the present application uses Monte Carlo method combined with S-G filtering algorithm to realize accurate and rapid estimation of the whole vehicle battery capacity.

[0062] 4、For the problem that preheating time is difficult to obtain, the present application constructs a preheating stage estimation model based on robust least square method, which greatly improves the accuracy of preheating time prediction.

[0063] The advantages of the additional aspects of the present application will be partially given in the following description, partially will become obvious from the following description, or be known by the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0064] The accompanying drawings, which form a part of this specification, are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification. The embodiments of the application, and their

[0065] Figure 1 A flowchart of an electric vehicle charging strategy recognition method provided for embodiment 1 of the application;

[0066] Figure 2 A schematic diagram of a feature table provided for embodiment 1 of the application;

[0067] Figure 3 A flowchart of an electric vehicle charging duration prediction method provided for embodiment 2 of the application. DETAILED DESCRIPTION

[0068] The application will be further described below with reference to the drawings and embodiments.

[0069] It should be noted that the following detailed description is exemplary in nature and is intended to provide further description of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0070] In the case of no conflict, the embodiments in the application and the features in the embodiments can be combined with each other.

[0071] Embodiment 1:

[0072] In different power charging devices, the charging conditions are quite different, and there is randomness in actual application. In a large amount of historical charging data, the charging conditions are complex and diverse, and the original charging strategy and various jump conditions are unknown. Therefore, the present implementation proposes an electric vehicle charging strategy recognition method, including the following processes:

[0073] S1: Obtain the real vehicle data set. The real vehicle data set of the present implementation includes time, charging state, total voltage, total current, SOC, minimum single cell temperature and other data.

[0074] S2: Data cleaning. The real vehicle data set has the following characteristics: large data size, with sample quantity reaching millions of rows; high data feature dimension, with large information quantity; many abnormal points, with missing key information. Based on this, the application reorders the time disorder data, uses Gaussian filtering for data smoothing processing to reduce noise, eliminates abnormal values, and uses linear interpolation to complete the missing data.

[0075] S3: Charging segment division. Due to unknown charging strategy details, only historical charging and discharging curves, actual working condition operation complexity, charging and discharging data mixed together, charging segment without label and unknown number. In order to analyze the charging strategy, the data needs to be divided into charging segments first, specifically, including:

[0076] The real vehicle data is screened, the data in charging state 1 (indicating charging) or 4 (indicating charging completion) is screened, the screened data is traversed, for two adjacent data, whether the absolute value of the SOC difference is greater than 1 and the time difference is greater than five sampling intervals, if not meet the above conditions, the two data points are regarded as the same charging segment, with the same charging segment ID; otherwise, the charging segment ID is increased by 1; after labeling all data points (charging segment ID), check the start and end SOC of each charging segment, and remove the charging segment with a start and end SOC difference less than 20% (i.e. a set threshold) because its information is too little to reflect the charging strategy characteristics.

[0077] S4: Charging strategy identification. The present application proposes a multi-level clustering-probability distribution collaborative charging strategy identification method based on machine learning, and the calculation process is as shown in Figure 1 .

[0078] Constructing a feature matrix: take it as input, the feature matrix is as follows:

[0079] X=[Time, U, I, SOC, T, ID] (1);

[0080] Wherein, X represents the feature matrix, Time represents the time, U represents the total voltage, I represents the total current, SOC represents the battery state of charge, T represents the minimum single cell temperature of the battery pack, and ID represents the charging segment ID.

[0081] Current gear first clustering: extract the current feature, use DBSCAN algorithm for current gear first clustering, DBSCAN algorithm as an unsupervised clustering algorithm, without pre-determining the number and center of clusters, the specific process is as follows:

[0082] a, select a point for inspection. Randomly select a point from the data set. If the point has been visited, select the next point.

[0083] b, determine whether it is a core point. Calculate the number of neighbors within the radius of the point. If the point meets the condition of the core point, a new cluster is formed with the point as the core, and the formula is as follows:

[0084] N ε (P)>MinPts (2);

[0085] Where, ε is the maximum distance of a point within the radius of ε neighbors, MinPts is the minimum number of neighbors required for a point to be a core point (including the point itself).

[0086] c. Extend the cluster. Extend all the points directly density reachable from this core point, and mark them as belonging to the same cluster. The definition of directly density reachable is: if point q is within the ε neighborhood of point p, and point p is a core point, then point q is a directly density reachable point of point p.

[0087] d. Handle border points and noise points. For points that cannot be extended (i.e. not a core point but within the ε neighborhood of a core point), mark them as border points. For points that are neither a core point nor within the ε neighborhood of any core point, mark them as noise points.

[0088] e. Calculate the cluster center of each cluster. The calculation formula is as follows:

[0089]

[0090] Where, C is the center of the cluster, n is the number of points in the cluster, I i is the i-th current value in the cluster.

[0091] f. Output the clustering result. Label each data point with a current gear, and output the cluster center current value of each current gear.

[0092] Generate a feature matrix: according to the feature Time, count the duration of each charging segment at different gears, and form a feature table as shown in Figure 2 For example, the charging segment with ID 1 has a current of 1909 seconds at gear 1, a current of 11412 seconds at gear 2, and no current at other gears.

[0093] Construct a new feature matrix D, which represents the duration of each charging segment at different gears:

[0094]

[0095] Where, i represents the total number of gears, j represents the total number of charging segment IDs, t represents the duration, and t d represents the duration of all gears for a charging segment.

[0096] Second-level clustering of charging segments: perform K-means clustering on the feature matrix D. The number of clusters can be determined artificially, thereby dividing typical and special charging strategies. The specific process is as follows:

[0097] A. Randomly select K initial centroids, where the value of K is 2;

[0098] B. For each charging segment j, calculate its Euclidean distance to each centroid and assign it to the cluster corresponding to the nearest centroid, formula as follows:

[0099]

[0100] where c k is the kth centroid, c ik is the value of the kth centroid in the i-th dimension.

[0101] C. Recalculate the centroid of each cluster, formula as follows:

[0102]

[0103] where C k is the set of charging segments in the kth cluster.

[0104] D. Repeat the steps of assigning clusters and updating centroids until the centroids no longer change or the maximum number of iterations is reached.

[0105] E. The result is output, each charging segment j will be assigned a cluster label k, indicating its belonging cluster.

[0106] Key point recognition: extract current data, perform least squares linear fitting, extract slope feature and mutation time, obtain minimum monomer temperature T data according to mutation time point and calculate probability distribution.

[0107] Finally, the data set is divided into typical charging strategy and special charging strategy, and the typical charging strategy can be determined as approximate two-stage constant current charging. The first stage is small current ramp charging and preheating, and when the minimum monomer temperature reaches 15 degrees Celsius (set temperature threshold), it enters the second stage of constant current charging, until the power is fully charged, and the subsequent embodiment 3 is based on this charging strategy to predict the charging duration. The number of charging segments using the special charging strategy is small, and the charging current fluctuates greatly, which does not have the general characteristics, so it does not need to be specially modeled for duration prediction.

[0108] Embodiment 2:

[0109] The present implementation provides an electric vehicle charging strategy recognition system, comprising:

[0110] The feature matrix construction unit is configured to: divide the charging segments according to the cleaned real vehicle data of the electric vehicle, and construct a feature matrix according to the charging segment division result and the cleaned real vehicle data of the electric vehicle;

[0111] The first-level clustering unit is configured to: extract the current features in the feature matrix, and perform first-level clustering of the current gear;

[0112] A new feature matrix construction unit is configured to: extract time features from the feature matrix, count the duration of each charging segment at different gears, and construct a new feature matrix for representing the duration of each charging segment at different gears;

[0113] The secondary clustering unit is configured to perform secondary clustering on the new feature matrix to obtain identification results of typical charging strategies and special charging strategies.

[0114] It is understandable that each of the above-mentioned units can be separately or completely combined into one or several other units to constitute, or one (or some) of the units can be further divided into multiple functionally smaller units to constitute, which can achieve the same operation without affecting the realization of the technical effects of the embodiments of the present application. The above-mentioned units are divided based on logical functions. In actual applications, the functions of one unit can also be implemented by multiple units, or the functions of multiple units can be implemented by one unit. In other embodiments of the present application, the electric vehicle charging strategy identification system may also include other units. In actual applications, these functions can also be implemented with the assistance of other units, and can be implemented by the collaboration of multiple units.

[0115] According to another embodiment of the present application, the system described in this embodiment can be constructed, and the electric vehicle charging strategy identification method of the embodiment of the present application can be implemented by running a computer program (including program code) capable of executing the steps involved in the corresponding method described in Example 1 on a general-purpose computing device such as a computer, which includes processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM). The computer program can be recorded on, for example, a computer-readable recording medium, and loaded into the above-mentioned computing device through the computer-readable recording medium and run therein.

[0116] Example 3:

[0117] like Figure 3 As shown, this implementation provides a method for predicting the charging time of an electric vehicle, which includes the following steps for the typical charging strategy identified in Example 1:

[0118] The training set and test set are divided into a ratio of 9:1, and then a charging time prediction model is constructed. The specific process is as follows:

[0119] Capacity estimation: To reduce the marginal error, select the longest current stable charging segment and use the inverse ampere-hour integration method to calculate the battery capacity of the current charging segment:

[0120]

[0121] where Q is the battery capacity, I(t) is the current at time t, SOC(t0) and SOC(t1) are the starting and ending SOC of the charging segment, t0 and t1 are the starting and ending time of the charging segment.

[0122] In order to reduce the error, combined with the idea of Monte Carlo, the continuous charging process is randomly cut from the current charging segment, cut 5 times, respectively calculate the capacity of the charging segment, select the mean value as the capacity of the current charging segment:

[0123]

[0124] where Q i is the battery capacity calculated each time, Q fin is the final battery capacity, and n is 5 here.

[0125] Add Savitzky-Golay filter, adopt robust least squares method, fit the capacity-time curve, and finally get the aging correction coefficient:

[0126] Q fin = a*Time + b (9);

[0127] where Time is time, and a and b are parameters to be fitted.

[0128] Preheating stage charging duration prediction: In cold weather, electric vehicle charging needs to be preheated. Considering the calculation complexity, combined with the linear relationship between temperature increment and preheating time, robust least squares method is used to establish the relationship between temperature increment and temperature increment time:

[0129] ΔT = T end - T0 (10);

[0130] t1 = a*ΔT + b (11);

[0131] where ΔT is the temperature increment, T0 and T end are the minimum temperature of the single body at the start of the preheating stage, t1 is the preheating stage time, and a and b are parameters to be fitted.

[0132] Second stage charging duration prediction: It can be seen that the starting charging current of the second stage and the starting SOC of the second stage present a linear relationship, and support vector regression method (SVR) is used to establish the relationship between charging SOC and current value:

[0133] I0 = ω*SOC0 + b (12);

[0134] where I0 is the starting current of the second stage, SOC0 is the starting SOC of the second stage, ω is the weight vector, and b is the bias term.

[0135] To train the SVR model, the following loss function needs to be minimized:

[0136]

[0137] where C is the regularization function, ε is the tolerance parameter, n is the number of charging segments involved in training, and i is the charging segment ID.

[0138] Since the charging duration needs to be calculated in real time, the charging current needs to be simplified to obtain the average charging current during the charging process. Only the initial current and the final current are averaged for simplification. The second stage duration prediction formula is

[0139]

[0140] where ΔSOC is the start-end SOC difference, I1 is the end current, and I2 is a constant.

[0141] Supplementary charging duration prediction: After the second stage charging is completed, the SOC reaches 99%, and a small current is needed for energy supplement. The charging time of this process is t3:

[0142]

[0143] where I end is the supplementary current.

[0144] Final duration prediction: add the prediction results of the preheating stage charging duration, the second stage charging duration, and the supplementary charging duration to obtain the complete charging duration:

[0145] t total = t1 + t2 + t3 (16);

[0146] where t total is the complete charging duration prediction result.

[0147] Example 4:

[0148] The present implementation provides an electric vehicle charging duration prediction system for predicting the duration of the typical charging strategy identified by the electric vehicle charging strategy identification method of Embodiment 1 of the present invention, comprising:

[0149] A first prediction unit configured to establish a temperature increment-temperature increment time relationship using robust least squares to obtain a preheating stage charging duration prediction result;

[0150] A second prediction unit configured to establish a charging SOC-current value relationship using support vector regression, combined with the battery capacity, to obtain a charging duration prediction result for the two-stage constant current charging stage;

[0151] The energy supplement time acquisition unit is configured to, after the end of the secondary constant current charging stage, when the SOC reaches 99%, supplement energy with a small current to obtain the energy supplement time;

[0152] The comprehensive prediction unit is configured to take the sum of the preheating stage charging time prediction result, the secondary constant current charging stage charging time prediction result, and the energy supplement time as the final charging time prediction result.

[0153] It can be understood that the above various units can be combined into one or several other units to constitute, or some of the units can be further split into a plurality of units with smaller functions to constitute, which can achieve the same operation without affecting the implementation of the technical effects of the embodiments of the present application. The above units are divided based on logical functions, and in actual application, the functions of one unit can also be implemented by multiple units, or the functions of multiple units can be implemented by one unit. In other embodiments of the present application, the electric vehicle charging time prediction system can also include other units, and in actual application, these functions can also be assisted by other units to achieve, and can be achieved by cooperation of multiple units.

[0154] According to another embodiment of the present application, the system described in the embodiment can be constructed and the electric vehicle charging time prediction method of the embodiments of the present application can be implemented by running a computer program (including program codes) capable of performing each step involved in the corresponding method described in Embodiment 3 on a general computing device such as a computer including processing elements and storage elements such as a central processing unit (CPU), a random access memory (RAM), a read-only memory (ROM), etc., the computer program can be recorded on a computer readable recording medium, and loaded into the above computing device through the computer readable recording medium and run therein.

[0155] Embodiment 5:

[0156] The present implementation provides an electronic device, which includes a processor, a communication interface, and a computer readable storage medium. Wherein the processor, the communication interface, and the computer readable storage medium are connected through a bus or other means.

[0157] Wherein, the communication interface is used to receive and send data, the computer readable storage medium can be stored in the memory of the electronic device, the computer readable storage medium is used to store computer programs, the computer programs include program instructions, and the processor is used to execute the program instructions stored in the computer readable storage medium.

[0158] The processor (or CPU (Central Processing Unit)) is the computing core and control core of the electronic device, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions to implement a corresponding method flow or a corresponding function.

[0159] The processor is configured to perform the process of embodiment 1 or embodiment 3.

[0160] Embodiment 6:

[0161] The present implementation provides a computer readable storage medium (Memory), which is a memory device in an electronic device, used to store programs and data. It can be understood that the computer readable storage medium here can include the built-in storage medium in the electronic device, and of course can also include the expansion storage medium supported by the electronic device. The computer readable storage medium provides a storage space, which stores the processing system of the electronic device.

[0162] And in the storage space, one or more instructions suitable for being loaded and executed by the processor are also stored, which can be one or more computer programs (including program codes). It should be noted that the computer readable storage medium here can be a high-speed RAM memory, or a non-volatile memory such as at least one disk memory; optionally, it can also be at least one computer readable storage medium located away from the aforementioned processor.

[0163] In one embodiment, the computer readable storage medium stores one or more instructions; the processor loads and executes the one or more instructions stored in the computer readable storage medium to implement the process of embodiment 1 or embodiment 3.

[0164] Embodiment 7:

[0165] The present implementation provides a computer program product or computer program, which includes computer instructions stored in a computer readable storage medium. The processor of the electronic device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the electronic device performs the process of embodiment 1 or embodiment 3.

[0166] Those skilled in the art can be aware that units and algorithm steps of each example described in combination with the embodiments disclosed in the application can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on specific applications and design constraints of the technical solutions. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the application.

[0167] In the above embodiments, all or part of the embodiments can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions according to the embodiments of the application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. Computer instructions can be stored in or transmitted by a computer-readable storage medium. Computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (for example, coaxial cable, optical fiber, digital line (DSL)) or wireless (for example, infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data processing device such as a server, data center, etc. containing one or more available media sets. The available media can be magnetic media (for example, floppy disk, hard disk, magnetic tape), optical media (for example, DVD), or semiconductor media (for example, solid state disk (SSD)) and the like.

[0168] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An electric vehicle charging strategy identification method, characterized by, The method comprises the following processes: charging segment division is performed according to the real vehicle data of the cleaned electric vehicle, a feature matrix is constructed according to the charging segment division result and the real vehicle data of the cleaned electric vehicle; current features in the feature matrix are extracted, and first-level clustering of current gears is performed; time features in the feature matrix are extracted, the duration of each charging segment at different gears is counted, and a new feature matrix for representing the duration of each charging segment at different gears is constructed; second-level clustering is performed on the new feature matrix, and then the recognition result of typical charging strategies and special charging strategies is obtained; the typical charging strategy is approximate two-stage constant current charging, the first stage is small current ramp charging and preheating, when the minimum cell temperature reaches a set temperature threshold, the second-stage constant current charging stage is entered, and the charging is completed until the power is full.

2. The electric vehicle charging strategy recognition method of claim 1, wherein the charging segment division performed according to the real vehicle data of the cleaned electric vehicle comprises: the real vehicle data after data cleaning is filtered, data in a charging state of charging or charging completion is filtered, the filtered data is traversed, for two adjacent data, it is judged whether the absolute value of the SOC difference is greater than 1 and the time difference is greater than five sampling intervals, if yes, the two adjacent data have the same charging segment ID; otherwise, the charging segment ID is increased by 1, the start and end SOC of each charging segment is checked, the charging segment with a start and end SOC difference less than a set threshold is removed, and finally the charging segment division result is obtained.

3. The electric vehicle charging strategy recognition method of claim 1 or 2, wherein the first-level clustering of current gears comprises: a data point is randomly selected from the data set, if the data point has been visited, the next data point is selected; for points that cannot be expanded, they are marked as boundary points, for points that are neither core points nor in the ε neighborhood of any core point, they are marked as noise points.

4. The electric vehicle charging strategy recognition method of claim 1 or 2, wherein the second-level clustering of the new feature matrix comprises: The number of neighbors within the radius ε of the selected data point is calculated, and if the data point satisfies the condition of the core point, a new cluster is formed with the data point as the core: wherein, is the maximum distance defining a neighbor of a point within the radius ε, is the minimum number of neighbors required for a point to be a core point; All points directly density reachable from this core point are expanded and marked as belonging to the same cluster; a point q is directly density reachable from a point p if point q is in the neighborhood of point p and point p is a core point. A point q is directly density reachable from a point p if point q is in the neighborhood of point p and point p is a core point. K initial centroids are randomly selected, where the value of K is 2, for dividing a typical charging strategy cluster and a special charging strategy cluster; calculating a cluster center for each cluster, wherein, is the center of the cluster, n is the number of points in the cluster, is the i-th current value in the cluster. i ​ the steps of assigning clusters and updating centroids are repeated until the centroids no longer change or the maximum number of iterations is reached.

5. The electric vehicle charging strategy recognition method of claim 4, wherein the total current features in the feature matrix are extracted, the total current features are linearly fitted by the least squares method, the slope feature and the mutation time are extracted, the minimum cell temperature data are obtained according to the mutation time point, and the probability distribution is calculated. It comprises: For each charging segment 𝑗, calculate its Euclidean distance to each centroid and assign it to the cluster corresponding to the nearest centroid: ,in, It is k centroid, No. k The centroid of i The values ​​in the dimensions; Recalculate the centroid of each cluster: where, is the set of charged segments in the k th cluster. a feature matrix construction unit configured to perform charging segment division according to the real vehicle data of the cleaned electric vehicle, and construct a feature matrix according to the charging segment division result and the real vehicle data of the cleaned electric vehicle; Each charging segment j is assigned a cluster label k , indicating the cluster to which the charging segment j belongs. a first-level clustering unit configured to extract current features in the feature matrix and perform first-level clustering of current gears; ​ 6. An electric vehicle charging strategy identification system characterized by, ​ ​ ​ The new feature matrix construction unit is configured to extract time features in the feature matrix, count the duration of each charging segment at different gears, and construct a new feature matrix for representing the duration of each charging segment at different gears. The secondary clustering unit is configured to perform secondary clustering on the new feature matrix, thereby obtaining an identification result of a typical charging strategy and a special charging strategy. The typical charging strategy is an approximate two-stage constant current charging, a first stage is a small current ramp charging and preheating, and when the minimum single cell temperature reaches a set temperature threshold, a second stage of constant current charging is entered until the power is fully charged.

7. An electric vehicle charging duration prediction method, characterized in that, The typical charging strategy identified by the electric vehicle charging strategy identification method of any one of claims 1-5 is subjected to duration prediction, including the following processes: A robust least squares method is used to establish a temperature increment-temperature increment time relationship to obtain a preheating stage charging duration prediction result; A support vector regression method is used to establish a charging SOC-current value relationship, combined with the battery capacity, to obtain a charging duration prediction result of the second constant current charging stage; After the end of the second constant current charging stage, when the SOC reaches 99%, a small current is used for energy supplement to obtain a supplement time; The sum of the preheating stage charging duration prediction result, the charging duration prediction result of the second constant current charging stage, and the supplement time is taken as the final charging duration prediction result.

8. The electric vehicle charging duration prediction method of claim 7, characterized in that, The robust least square method is used to establish the temperature increment-temperature increment time relationship to obtain the preheating stage charging time prediction results, including: , is the temperature increment, and are the minimum temperatures of the single body at the start of the preheating stage, is the preheating stage time, and are the parameters to be fitted; A support vector regression method is used to establish the relationship between the charging SOC and the current value: wherein, is the second stage starting current, is the second stage starting SOC, is a weight vector, is a bias term; The charging duration prediction result of the secondary constant current charging stage is: wherein, is the start-stop SOC difference value; is the end current, which is a constant; is the battery capacity; The energy supplement time is wherein, is the energy supplement current.

9. The electric vehicle charging duration prediction method of claim 8, characterized in that, The battery capacity acquiring, comprising: The longest current stable charging segment is selected, and the current charging segment battery capacity is calculated by using the inverse ampere-hour integration method: wherein, is the battery capacity, is the time t current, and are the start and end SOC of the charging segment, and are the start and end time of the charging segment; A continuous charging process is randomly intercepted from the current charging segment, and the intercepted n The capacity of the charging segment is calculated, and the average value is selected as the capacity of the current charging segment: wherein, is the battery capacity calculated each time, is the final battery capacity; The Savitzky-Golay filter is added, a robust least squares method is used to fit the capacity-time curve, and the following is obtained: wherein, is time, and is the parameter to be fitted.

10. An electric vehicle charging duration prediction system, characterized by, The typical charging strategy identified by the electric vehicle charging strategy identification method of any one of claims 1-5 is subjected to duration prediction, including: The first prediction unit is configured to use a robust least squares method to establish a temperature increment-temperature increment time relationship to obtain a preheating stage charging duration prediction result; The second prediction unit is configured to use a support vector regression method to establish a charging SOC-current value relationship, combined with the battery capacity, to obtain a charging duration prediction result of the second constant current charging stage; The supplement time acquisition unit is configured to use a small current for energy supplement after the end of the second constant current charging stage when the SOC reaches 99% to obtain a supplement time; The comprehensive prediction unit is configured to take the sum of the preheating stage charging duration prediction result, the charging duration prediction result of the second constant current charging stage, and the supplement time as the final charging duration prediction result.

Images