Characteristic extraction method, system and device for charging behavior of electric vehicle, and medium

Abstract

The invention discloses an electric vehicle charging behavior feature extraction method, system and device and a medium. The method comprises the steps that a charging behavior original sample set C is acquired; classifying the charging behavior original sample set C by using a trained random forest model to obtain a leaf node index matrix P to which each sample in the charging behavior original sample set C falls in all decision trees of the forest; performing Gaussian hybrid clustering on a leaf node index matrix P to which each sample in the charging behavior original sample set C falls in all decision trees of the forest to obtain a clustering result; and carrying out feature extraction on the charging behavior of the charged vehicle from the clustering result. The method, the system, the equipment and the medium can accurately carry out feature extraction on the charging behavior of the electric vehicle.

Description

Technical Field

[0001] The present invention belongs to the technical field of interaction between electric vehicles and power grids, and relates to a method, system, equipment and medium for extracting characteristics of charging behavior of electric vehicles. Background Art

[0002] Due to the explosive growth of electric vehicle ownership, the spatiotemporal uncertainty of charging demand, and the lack of coordinated planning with the automobile, transportation, and energy industries during the construction and transformation of charging and swapping facilities, the phenomenon of "cars without charging piles, charging piles without cars" is common, and the rational planning and optimal allocation of charging resources are imperative. There are significant differences in vehicle configuration, behavioral habits, charging scenarios and other factors among different electric vehicle users, which leads to the randomness and user heterogeneity of the spatiotemporal distribution of charging demand. Therefore, it is necessary to further study the potential patterns and laws of user charging behavior based on the historical behavior data of electric vehicles in order to more finely and accurately predict the charging load of electric vehicles, that is, to extract features of electric vehicle charging behavior.

[0003] Most existing studies on electric vehicle charging behavior tend to simplify the classification based on a single dimension such as vehicle type, charging location or charging power. Building models based on subjective assumptions may affect the accuracy and universality of the models. In addition, the data sets used in most studies have a small sample size, cover few vehicle models or have limited application scenarios, making it difficult to conduct an in-depth and comprehensive analysis of charging behavior. Therefore, it is difficult to extract features of electric vehicle charging behavior. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a method, system, device and medium for extracting the characteristics of electric vehicle charging behavior. The method, system, device and medium can accurately extract the characteristics of electric vehicle charging behavior.

[0005] In order to achieve the above object, the present invention adopts the following technical scheme:

[0006] In one aspect of the present invention, the method for extracting features of electric vehicle charging behavior of the present invention comprises:

[0007] Obtaining a charging behavior original sample set C;

[0008] Using the trained random forest model to classify the charging behavior original sample set C, obtaining a leaf node index matrix P of each sample in the charging behavior original sample set C in all decision trees in the forest;

[0009] Performing Gaussian mixture clustering on the leaf node index matrix P where each sample in the original sample set C of the charging behavior falls in all decision trees in the forest to obtain a clustering result;

[0010] The feature extraction of the charging behavior of the charging vehicle is performed from the clustering result.

[0011] The feature extraction method of electric vehicle charging behavior of the present invention is further improved in that:

[0012] Furthermore, the method of using the trained random forest model to analyze the charging behavior original sample set C also includes:

[0013] Select a subset of features;

[0014] Constructing a training sample set based on the feature subset;

[0015] Build a random forest model;

[0016] The random forest model is trained based on the training sample set to obtain a trained random forest model.

[0017] Furthermore, the process of selecting a feature subset is:

[0018] Calculate the contribution of all feature quantities of each sample in the original charging behavior sample set C to the accuracy of the random forest model;

[0019] According to the contribution of all feature quantities of each sample to the accuracy of the random forest model, the original sample set C of the charging behavior is sorted by feature importance, feature selection is performed according to the feature importance sorting result, and a feature subset is constructed according to the selected features.

[0020] Furthermore, all characteristic quantities of each sample include but are not limited to: charging start time, charging end time, charging duration, average power, charging amount, starting SOC, ending SOC, charging location and month.

[0021] Furthermore, feature x j Importance Rating for:

[0022]

[0023] Among them, PE(x j ) is the feature x j The mean change in the prediction error rate is , and M is the total number of features.

[0024] Furthermore, the process of extracting the features of the charging behavior of the charging vehicle from the clustering results is as follows:

[0025] The spatiotemporal distribution characteristics of the travel behavior of each vehicle type and the characteristics of each type of charging behavior are extracted from the clustering results.

[0026] Furthermore, the spatiotemporal distribution characteristics of the travel behavior of each vehicle type include the number of daily trips, travel demand time-OD set, trip start time, trip start SOC and parking time after driving; the characteristics of charging behavior include the joint distribution of start and end SOC and average power.

[0027] In a second aspect of the present invention, the method for extracting features of electric vehicle charging behavior of the present invention comprises:

[0028] An acquisition module is used to acquire an original sample set C of charging behavior;

[0029] A classification module, used to classify the charging behavior original sample set C using the trained random forest model, and obtain a leaf node index matrix P where each sample in the charging behavior original sample set C falls in all decision trees in the forest;

[0030] A clustering module, used for performing Gaussian mixture clustering on the leaf node index matrix P where each sample in the original sample set C of the charging behavior falls in all decision trees in the forest, to obtain a clustering result;

[0031] The feature extraction module is used to extract the features of the charging behavior of the charging vehicle from the clustering results.

[0032] In a third aspect of the present invention, the computer device of the present invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the feature extraction method of the electric vehicle charging behavior when executing the computer program.

[0033] In a fourth aspect of the present invention, the computer-readable storage medium of the present invention stores a computer program, and when the computer program is executed by a processor, the steps of the feature extraction method of the electric vehicle charging behavior are implemented.

[0034] The present invention has the following beneficial effects:

[0035] The feature extraction method, system, device and medium of the electric vehicle charging behavior described in the present invention use the neighbor information of the sample in the random forest to cluster the data during the specific operation, and can handle non-convex clusters and noise points. In addition, since the random forest can handle high-dimensional data and features with complex relationships, it may be more effective than the traditional clustering method in clustering high-dimensional data. The present invention uses the trained random forest model to classify the original sample set C of the charging behavior to improve the effectiveness of clustering, and at the same time performs Gaussian mixture clustering on the leaf node index matrix P, wherein the method of using the leaf node index matrix to characterize the sample similarity significantly reduces the computer memory and time spent in the calculation compared to the conventional dissimilarity matrix. It should be noted that the present invention uses an unfiltered large-scale electric vehicle behavior data set to carry out research, and for multiple types of electric vehicles, restores the complexity of their behavior patterns based on data mining technology, providing effective data support for refined demand forecasting. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0037] Figure 1 is a flow chart of the method of the present invention;

[0038] Figure 2 It is a system structure diagram of the present invention. DETAILED DESCRIPTION

[0039] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only an embodiment of a part of the present invention, not all embodiments, and is not intended to limit the scope of the present invention. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concepts disclosed in the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present invention.

[0040] The accompanying drawings show schematic diagrams of structures according to embodiments disclosed in the present invention. These figures are not drawn to scale, and some details are magnified and some details may be omitted for the purpose of clear expression. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0041] Embodiment 1

[0042] refer to Figure 1 The feature extraction method of electric vehicle charging behavior of the present invention comprises the following steps:

[0043] 1) Using the preprocessed charging behavior original sample set C to train the random forest model;

[0044] Among them, the preprocessing of electric vehicle charging behavior includes the preprocessing of the original data set and the preprocessing of each behavior data set.

[0045] The preprocessing process is as follows: first, the original data set containing the three types of electric vehicle behaviors is cleaned, and then for data of different state types, targeted abnormal data detection methods are adopted according to the state characteristics.

[0046] The preprocessing of the original data set includes data deduplication, deletion of redundant data, format conversion, missing value processing, illegal data screening, vehicle information supplementation, and grouping and sorting.

[0047] The specific steps for training a random forest model are:

[0048] 11) Based on the distribution of each parameter in the original charging behavior sample set C, a fake data set F with the same number of samples as the original charging behavior sample set C is randomly generated, wherein the generation of each parameter in the fake data set F is independent, and the original charging behavior sample set C and the fake data set F are merged to obtain a set At the same time, a label column l is added to the set T, in which the labels of the samples from the original charging behavior sample set C are assigned to 1, and the labels of the samples from the fake data set F are assigned to 0.

[0049] 12) Use the bootstrap method to perform random sampling with replacement on the set T, and construct the training set D1, D2, ..., D of each decision tree based on the results of random sampling. q , where q is the number of decision trees in the forest, TD i is the out-of-bag data (OOB) of the i-th tree.

[0050] 13) When any node in the decision tree needs to be split, m attributes are randomly selected from the M attributes of the sample without replacement, where the number of elements m in the feature subset satisfies Based on the Gini coefficient, the best splitting method for each node is obtained. The smaller the Gini coefficient, the better the classification effect.

[0051] 14) Repeat the node splitting process until the preset end condition is reached, and recursively generate each decision tree. No pruning operation is performed during the construction of the decision tree.

[0052] 2) Calculate the contribution of all sample features to the model accuracy, and perform feature importance ranking and feature selection on the original charging behavior sample set C;

[0053] Among them, all the characteristic quantities of the sample include charging start time, charging end time, charging duration, average power, charging amount, starting SOC, ending SOC, charging location and month.

[0054] The specific steps for feature importance sorting and feature selection of the original charging behavior sample set C are as follows:

[0055] 21) For decision tree i, calculate the original out-of-bag data OOB i The out-of-bag prediction error Err i1 ;

[0056] 22) OOB i x in j Add random noise to the column or perform random sequence permutation, keep other feature columns unchanged, and recalculate the prediction error Err for the new out-of-bag data i2 ;

[0057] 23) Perform steps 21) and 22) on the q trees in the forest respectively to obtain the feature x j The average change of the forecast error rate PE(x j ), where PE(x j )for:

[0058]

[0059] 24) Repeat steps 21) to 23) to obtain the mean of the error rate changes of all feature quantities, and perform normalization on the mean of the error rate changes of all feature quantities. is the characteristic quantity x j The importance rating of for:

[0060]

[0061] 25) Set a feature selection importance threshold, and select all feature quantities x according to the feature selection importance threshold j Filtering is performed to complete the feature subset selection process.

[0062] 3) Retrain the random forest model using the sample set containing the feature subset, and use the trained random forest model to classify and predict the original charging behavior sample set C, and obtain the leaf node index matrix P of each sample in the original charging behavior sample set C in all decision trees in the forest;

[0063] 4) Use out-of-bag samples to calculate the accuracy of classification prediction and evaluate the adaptability of the random forest model to the charging behavior sample set;

[0064] 5) Perform a Gaussian mixture clustering process on the index matrix P of each leaf node to obtain the final clustering result of the original charging behavior sample set C.

[0065] Among them, after determining the number of Gaussian components during clustering, the maximum expectation algorithm is used to solve the model parameters. The algorithm continuously iterates between the two steps of seeking expectation and seeking maximization. When the algorithm reaches convergence, the iterative result at least approaches the local maximum.

[0066] The specific steps of Gaussian mixture clustering are:

[0067] 51) Input data set x = {x1, x2, …, x m} and the number of components K of GMM, and randomly initialize the distribution parameters α of each component k , μ k and Σ k , where k∈{1,2,…,K};

[0068] 52) The method for calculating the posterior probability of each sample being generated by each mixed component and the posterior probability of the sample point xj being generated by the kth component is:

[0069]

[0070] Among them, j∈{1,2,…,m}.

[0071] 53) Update α according to the result of step 52) and the definition of mean and covariance matrix k , μ k and Σ k , the update formula is:

[0072]

[0073] 54) Repeat step 52) and step 53) until the random forest model converges or reaches the maximum number of iterations;

[0074] 55) For each sample point, calculate the posterior probability of the sample point under each component, and classify the sample point into the cluster represented by the component with the largest posterior probability. The calculation method is:

[0075]

[0076] in, Represents the sample point x j The cluster to which it belongs.

[0077] 6) Extract the spatiotemporal distribution characteristics of each vehicle type’s travel behavior and the characteristics of each type of charging behavior from the clustering results.

[0078] Among them, the spatiotemporal distribution characteristics of travel behavior of each model include the number of daily trips, travel demand time-OD set, trip start time, trip start SOC and parking time after driving; the charging behavior characteristics of electric vehicles include the joint distribution of start and end SOC and average power.

[0079] Embodiment 2

[0080] refer to Figure 2 The feature extraction system of electric vehicle charging behavior of the present invention comprises:

[0081] An acquisition module is used to acquire an original sample set C of charging behavior;

[0082] A classification module, used to classify the charging behavior original sample set C using the trained random forest model, and obtain a leaf node index matrix P where each sample in the charging behavior original sample set C falls in all decision trees in the forest;

[0083] A clustering module, used for performing Gaussian mixture clustering on the leaf node index matrix P where each sample in the original sample set C of the charging behavior falls in all decision trees in the forest, to obtain a clustering result;

[0084] The feature extraction module is used to extract the features of the charging behavior of the charging vehicle from the clustering results.

[0085] As an implementation manner of the present invention, before the charging behavior original sample set C is processed using the trained random forest model, the method further includes:

[0086] Select a subset of features;

[0087] Constructing a training sample set based on the feature subset;

[0088] Build a random forest model;

[0089] The random forest model is trained based on the training sample set to obtain a trained random forest model.

[0090] As an implementation mode of the present invention, the process of selecting a feature subset is:

[0091] Calculate the contribution of all feature quantities of each sample in the original charging behavior sample set C to the accuracy of the random forest model;

[0092] According to the contribution of all feature quantities of each sample to the accuracy of the random forest model, the original sample set C of the charging behavior is sorted by feature importance, feature selection is performed according to the feature importance sorting result, and a feature subset is constructed according to the selected features.

[0093] As an implementation mode of the present invention, all characteristic quantities of each sample include but are not limited to: charging start time, charging end time, charging duration, average power, charging amount, starting SOC, ending SOC, charging location and month.

[0094] As an embodiment of the present invention, feature x j Importance Rating for:

[0095]

[0096] Among them, PE(x j ) is the feature x j The mean change in the prediction error rate is , and M is the total number of features.

[0097] As an implementation mode of the present invention, the process of extracting the features of the charging behavior of the charging vehicle from the clustering results is as follows:

[0098] The spatiotemporal distribution characteristics of the travel behavior of each vehicle type and the characteristics of each type of charging behavior are extracted from the clustering results.

[0099] As an implementation mode of the present invention, the spatiotemporal distribution characteristics of the travel behavior of each vehicle model include the number of daily trips, travel demand time-OD set, trip start time, trip start SOC and parking time after driving; the characteristics of charging behavior include the joint distribution of start and end SOC and the average power.

[0100] The division of modules in the embodiments of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present application may be integrated into a processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0101] Embodiment 3

[0102] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of implementing the feature extraction method of the charging behavior of the electric vehicle are implemented, for example, including: obtaining a charging behavior original sample set C; using a trained random forest model to classify the charging behavior original sample set C, and obtaining a leaf node index matrix P of each sample in the charging behavior original sample set C in all decision trees in the forest; performing Gaussian mixture clustering on the leaf node index matrix P of each sample in the charging behavior original sample set C in all decision trees in the forest, and obtaining a clustering result; and extracting the features of the charging behavior of the charging vehicle from the clustering result. The memory may include a memory, such as a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk memory, etc. The processor, the network interface, and the memory are interconnected through an internal bus, and the internal bus may be an industrial standard architecture bus, a peripheral component interconnection standard bus, an extended industrial standard architecture bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The memory is used to store programs. Specifically, the program may include a program code, and the program code includes a computer operation instruction. The memory may include memory and nonvolatile memory and provides instructions and data to the processor.

[0103] Embodiment 4

[0104] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of implementing the feature extraction method of the charging behavior of the electric vehicle include: obtaining a charging behavior original sample set C; using a trained random forest model to classify the charging behavior original sample set C, and obtaining a leaf node index matrix P where each sample in the charging behavior original sample set C falls in all decision trees in the forest; performing Gaussian mixture clustering on the leaf node index matrix P where each sample in the charging behavior original sample set C falls in all decision trees in the forest, and obtaining a clustering result; and extracting the features of the charging behavior of the charging vehicle from the clustering result. Specifically, the computer-readable storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. The volatile memory may include a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may include a read-only memory (ROM), a hard disk, a flash memory, an optical disk, a magnetic disk, etc.

[0105] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0106] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0107] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0108] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for extracting features of electric vehicle charging behavior, characterized in that: include: Obtaining a charging behavior original sample set C; Using the trained random forest model to classify the charging behavior original sample set C, obtaining a leaf node index matrix P of each sample in the charging behavior original sample set C in all decision trees in the forest; Performing Gaussian mixture clustering on the leaf node index matrix P where each sample in the original sample set C of the charging behavior falls in all decision trees in the forest to obtain a clustering result; The feature extraction of the charging behavior of the charging vehicle is performed from the clustering result.

2. The feature extraction method of electric vehicle charging behavior according to claim 1 is characterized in that: Before the charging behavior original sample set C is processed using the trained random forest model, the following steps are further included: Select a subset of features; Constructing a training sample set based on the feature subset; Build a random forest model; The random forest model is trained based on the training sample set to obtain a trained random forest model.

3. The feature extraction method of electric vehicle charging behavior according to claim 2 is characterized in that: The process of selecting a feature subset is as follows: Calculate the contribution of all feature quantities of each sample in the original charging behavior sample set C to the accuracy of the random forest model; According to the contribution of all feature quantities of the samples to the accuracy of the random forest model, the original sample set C of the charging behavior is sorted by feature importance, feature selection is performed according to the feature importance sorting result, and a feature subset is constructed according to the selected features.

4. The feature extraction method of electric vehicle charging behavior according to claim 3 is characterized in that: All characteristic quantities of each sample include but are not limited to: charging start time, charging end time, charging duration, average power, charging amount, starting SOC, ending SOC, charging location and month.

5. The feature extraction method of electric vehicle charging behavior according to claim 1 is characterized in that: Features j Importance Rating for: Among them, PE(x j ) is the feature x j The mean change in the prediction error rate is , and M is the total number of features.

6. The feature extraction method of electric vehicle charging behavior according to claim 1 is characterized in that: The process of extracting the features of the charging behavior of the charging vehicle from the clustering results is as follows: The spatiotemporal distribution characteristics of the travel behavior of each vehicle type and the characteristics of each type of charging behavior are extracted from the clustering results.

7. The feature extraction method of electric vehicle charging behavior according to claim 6 is characterized in that: The spatiotemporal distribution characteristics of the travel behavior of each vehicle type include the number of daily trips, travel demand time-OD set, trip start time, trip start SOC and parking time after driving; the characteristics of the charging behavior include the joint distribution of start and end SOC and average power.

8. A method for extracting features of electric vehicle charging behavior, characterized in that: include: An acquisition module is used to acquire an original sample set C of charging behavior; A classification module, used to classify the charging behavior original sample set C using the trained random forest model, and obtain a leaf node index matrix P where each sample in the charging behavior original sample set C falls in all decision trees in the forest; A clustering module, used for performing Gaussian mixture clustering on the leaf node index matrix P where each sample in the original sample set C of the charging behavior falls in all decision trees in the forest, to obtain a clustering result; The feature extraction module is used to extract the features of the charging behavior of the charging vehicle from the clustering results.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the feature extraction method for electric vehicle charging behavior as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for extracting features of charging behavior of an electric vehicle as described in any one of claims 1 to 7 are implemented.

Images