Charging index evaluation method, decision tree model training method and storage medium

Abstract

The invention provides a charging index evaluation method, a decision tree model training method and a storage medium. The charging index evaluation method is applied to a cloud platform, a decision tree model is deployed in the cloud platform, and the method comprises the steps that a charging signal sent by a vehicle is received, and the charging signal comprises charging data; the charging data are input into a decision tree model to obtain a predicted value output by the decision tree model, and the predicted value is used for representing an output current capability value of a charging pile for supplying power to the vehicle; and sending the predicted value to the vehicle, so that the vehicle evaluates the charging index according to the predicted value. According to the method, the problem that errors exist in the output current capacity value of the charging pile obtained by the vehicle from the charging pile is avoided, so that the precision of the charging index for vehicle evaluation is improved, the charging experience is improved, and the charging safety is effectively guaranteed.

Description

Technical Field

[0001] The present application relates to the field of automobile charging, and specifically to a charging indicator evaluation method, a decision tree model training method and a storage medium. Background Art

[0002] A charging pile is an efficient vehicle charging device that can charge various types of electric vehicles according to different voltage levels. At present, when a vehicle is charged through a charging pile, the vehicle needs to obtain the output current capacity value of the charging pile through the controller area network (CAN) communication between the charging pile and the vehicle after the charging gun is plugged in, and the charging index is evaluated based on the output current capacity value of the charging pile.

[0003] However, due to various reasons, the charging pile itself may have some problems, resulting in a large error between the output current capacity value sent by the charging pile to the vehicle and the actual output current capacity value of the charging pile. As a result, the accuracy of the charging index evaluated by the electric vehicle based on the output current capacity value sent by the charging pile to the vehicle is very low, which affects the user's charging experience and makes it impossible to guarantee charging safety. Summary of the invention

[0004] The present application provides a charging indicator evaluation method, a decision tree model training method and a storage medium. The following introduces various aspects involved in the embodiments of the present application.

[0005] In a first aspect, a method for evaluating a charging index is provided, the method being applied to a cloud platform, wherein a decision tree model is deployed in the cloud platform, the method comprising: receiving a charging signal sent by a vehicle, the charging signal comprising charging data; inputting the charging data into the decision tree model to obtain a predicted value output by the decision tree model, the predicted value being used to characterize an output current capacity value of a charging pile supplying power to the vehicle; and sending the predicted value to the vehicle, so that the vehicle evaluates the charging index according to the predicted value.

[0006] Optionally, the charging data includes location information, and inputting the charging data into the decision tree model to obtain the predicted value output by the decision tree model includes: inputting the charging data into the decision tree model so that the decision tree model performs the following operations: determining the map-attached information of the charging pile based on the location information, the map-attached information of the charging pile including any one or more of the following data: the type of the charging pile, the density of commercial facilities within a preset distance of the charging pile, the density of residential areas within a preset distance of the charging pile, and the establishment time of the charging pile; outputting the predicted value based on the map-attached information of the charging pile and the charging data.

[0007] Optionally, the map-attached information of the charging pile includes the type of the charging pile and the density of residential areas within a preset distance of the charging pile, the charging data includes vehicle type information, and outputting the predicted value based on the map-attached information of the charging pile and the charging data includes: determining a first target output path based on the type of the charging pile, the first target output path being one of a plurality of first output paths, and the plurality of first output paths corresponding one-to-one to groups of types of a plurality of charging piles; in response to the first target output path indicating that the predicted value needs to be output, outputting the predicted value from the first target output path; in response to the first target output path indicating that a second target output path needs to be determined, determining a second target output path based on the density of residential areas and the first target output path The second target output path, the second target output path is one of a plurality of second output paths, and the plurality of second output paths correspond one-to-one to a plurality of groups of residential area density; in response to the second target output path indicating the need to output the predicted value, the predicted value is output from the second target path; in response to the second target output path indicating the need to determine a third target output path, the third target output path is determined according to the vehicle model information and the second target output path, the third target output path is one of a plurality of third output paths, and the plurality of third output paths correspond one-to-one to a plurality of groups of vehicle model information; in response to the third target output path needing to output the predicted value, the predicted value is output from the third target output path.

[0008] Optionally, the decision tree model is trained based on training data, and the training data is obtained based on vehicle charging behavior data and a clustering algorithm.

[0009] In a second aspect, a method for evaluating a charging index is provided, which is applied to a vehicle, and the method comprises: sending a charging signal to a cloud platform, wherein the charging signal comprises charging data; receiving a predicted value sent by the cloud platform, wherein the predicted value is used to characterize an output current capacity value of a charging pile supplying power to the vehicle and the predicted value is obtained by a decision tree model deployed in the cloud platform based on the charging data; and evaluating the charging index according to the predicted value.

[0010] In a third aspect, a method for training a decision tree model is provided, the method being used to train the decision tree model as described in the first aspect, the method comprising: obtaining charging behavior data of a vehicle; determining training data based on the charging behavior data of the vehicle and a clustering algorithm; training the decision tree model based on the training data; and updating the parameters of the decision tree model based on the output of the decision tree model.

[0011] Optionally, the charging behavior data of the vehicle includes location information, and determining the training data based on the charging behavior data of the vehicle and the clustering algorithm includes: clustering the charging behavior data of the vehicle according to the clustering algorithm and the location information to obtain aggregate information of the charging pile, the aggregate information of the charging pile including a location tag; acquiring map-attached information of the charging pile according to the location tag; and integrating the aggregate information of the charging pile and the map-attached information of the charging pile into the training data.

[0012] Optionally, determining the training data based on the charging behavior data of the vehicle and the clustering algorithm includes: preprocessing the charging behavior data of the vehicle to obtain processed charging behavior data of the vehicle; and determining the training data based on the processed charging behavior data of the vehicle and the clustering algorithm.

[0013] Optionally, obtaining the map-attached information of the charging pile according to the location tag includes: obtaining the map-attached information of the charging pile through a map application programming interface API interface according to the location tag, the map-attached information of the charging pile including any one or more of the following data: the type of the charging pile, the density of commercial facilities within a preset distance of the charging pile, the density of residential areas within a preset distance of the charging pile, and the establishment time of the charging pile.

[0014] According to a fourth aspect, a computer-readable storage medium is provided, wherein the computer storage medium stores a computer program, and when the computer program is executed, the method according to the first aspect, the second aspect, or the third aspect is implemented.

[0015] The charging index evaluation method provided in the embodiment of the present application, in which the vehicle can interact with a cloud platform deployed with a decision tree model, wherein the decision tree model can automatically predict the output current capacity value of the charging pile for charging the vehicle based on the charging data in the charging signal sent by the vehicle, and the vehicle can evaluate the charging index based on the predicted value output by the decision tree model. In this way, the problem of errors in the output current capacity value of the charging pile obtained by the vehicle from the charging pile can be avoided, thereby improving the accuracy of the charging index evaluated by the vehicle, improving the charging experience and effectively ensuring charging safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flowchart of a method for training a decision tree model provided in an embodiment of the present application.

[0017] Figure 2 A flow chart of a charging indicator evaluation method provided in an embodiment of the present application.

[0018] Figure 3A flowchart of a method for outputting a prediction value using a decision tree model provided in an embodiment of the present application.

[0019] Figure 4 A flowchart of a method for training and using a decision tree model provided in an embodiment of the present application.

[0020] Figure 5 A flowchart of the DBSCAN algorithm provided in an embodiment of the present application.

[0021] Figure 6 A schematic diagram of the calibration of a cluster point set obtained using the DBSCAN algorithm provided in an embodiment of the present application. DETAILED DESCRIPTION

[0022] In order to facilitate understanding of the present application, the present application is described in more detail below based on exemplary embodiments and in conjunction with the accompanying drawings. The same or similar reference numerals are used in the accompanying drawings to represent the same or similar modules. It should be understood that the accompanying drawings are only exemplary and the scope of protection of the present application is not limited thereto.

[0023] Due to its environmentally friendly characteristics, the new energy vehicle industry has shown a rapid development trend in recent years. At present, the global sales of electric vehicles have exceeded 12 million, an increase of about 50% year-on-year. Correspondingly, the construction and development of charging infrastructure has also developed rapidly. In 2024, the cumulative number of charging infrastructure nationwide reached 11.433 million units, an increase of 49.6% over the same period last year. It can be seen that the construction of charging infrastructure has developed rapidly this year. Charging infrastructure includes charging piles (or charging stations), which are efficient vehicle charging devices that can charge various models of electric vehicles according to different voltage levels.

[0024] At present, when a vehicle is charged by a charging pile, the vehicle needs to obtain the output current capability value of the charging pile (for example, the maximum output current capability value of the charging pile) through the controller area network (CAN) communication between the charging pile and the vehicle (for example, the battery management system (BMS) in the vehicle) after the charging gun is plugged in, and evaluate the charging indicators (for example, the remaining charging time, the state of charge (SOC), the battery temperature, etc.) according to the output current capability value of the charging pile.

[0025] However, due to the rapid construction of charging infrastructure, some problems with charging piles have emerged one after another. For example, due to the large number of charging pile brands on the market, the technical level and service quality of each charging pile vary. For another example, some charging piles may have design defects, resulting in low charging efficiency. For another example, some charging piles have a high failure rate due to improper maintenance. There are even some charging piles that may have safety hazards, such as poor contact of the charging interface and insufficient protection level. These problems directly affect the charging experience and safety of new energy vehicle users.

[0026] Due to the above problems, during the charging process, the output current capacity value of the charging pile that the charging pile interacts with the new energy vehicle is inconsistent with the output current capacity value of the charging pile during the actual charging process. According to the statistics of the latest big data, this inconsistency accounts for as much as 68.5% of the total charging behavior. For example, more than 50% of the maximum output current capacity values ​​of 400A charging piles can only reach a maximum current of 200A or even 150A during the actual charging process. The actual capacity value is less than half of the predicted capacity. In other words, the output current capacity value of the charging pile interacted with the charging pile is likely to have an error with the actual value. This causes the vehicle to misjudge the maximum output current capacity value of the charging pile when evaluating the charging index, resulting in low accuracy of the charging index evaluated by the vehicle. Affects the user's charging experience, and even affects charging safety over time.

[0027] In order to solve the above problems, the embodiment of the present application provides a charging index evaluation method, in which a vehicle can interact with a cloud platform deployed with a decision tree model, wherein the decision tree model can automatically predict the output current capacity value of the charging pile for charging the vehicle based on the charging data in the charging signal sent by the vehicle, and the vehicle can evaluate the charging index based on the predicted value output by the decision tree model. This method can avoid the problem of errors in the output current capacity value of the charging pile obtained by the vehicle from the charging pile, thereby improving the accuracy of the charging index evaluated by the vehicle, improving the charging experience and effectively ensuring charging safety.

[0028] As mentioned above, the embodiment of the present application uses a decision tree model. The decision tree model is a machine learning model obtained by supervised learning, and the model is good at classification and regression tasks. The decision tree model represents the decision rule by constructing a tree structure, each internal node represents a test on an attribute, each branch represents a test result, and each leaf node represents a category or predicted value. Based on this, the decision tree model includes a root node, an internal node, a branch (or path), and a leaf node.

[0029] The decision tree model works by recursively dividing the data set into smaller and smaller subsets until all samples in the subsets belong to the same class or a certain stopping condition is reached. Each time a split is made, the best attribute is selected to split in order to maximize information gain or minimize impurity.

[0030] In an embodiment of the present application, a decision tree model is used to output a predicted value for characterizing the output current capability value of a charging pile that supplies power to a vehicle based on charging data in a charging signal. That is, the leaf nodes of the decision tree model in an embodiment of the present application are used to output predicted values. Preferably, the predicted value includes the maximum output current capability value of the charging pile that supplies power to the vehicle. The maximum output current capability value of the charging pile is a key parameter used to evaluate charging indicators in the following text. By including the maximum output current capability value of the charging pile in the predicted value, the accuracy of the charging indicators evaluated in the following text can be further improved.

[0031] In order to facilitate the understanding of the decision tree model in the embodiment of the present application, Figure 1 The training method of the decision tree model provided in the embodiment of the present application is introduced in detail. It should be noted that the training process of the decision tree model can be performed by the cloud platform described later, or the training process of the decision tree model can be performed by another cloud platform or server different from the cloud platform described later.

[0032] like Figure 1 As shown, the training method of the decision tree model provided in the embodiment of the present application includes steps S110-S140.

[0033] In step S110, the charging behavior data of the vehicle is acquired.

[0034] The vehicle charging behavior data includes a large amount of vehicle charging behavior data. The vehicle charging behavior data can be reported to the cloud platform by all vehicles that can interact with the cloud platform described later each time they are charged. The vehicle charging behavior data includes, but is not limited to, any one or more of the following information: charging location information (e.g., longitude information of the charging location and latitude information of the charging location), the maximum actual current of the charging process, the maximum limit current of the vehicle charging process, vehicle model information, battery type, charging time, etc.

[0035] In some embodiments, the charging behavior data of the vehicle may include but is not limited to the charging location information, the maximum actual current of the charging process, the maximum limit current of the vehicle charging process, the vehicle model information, the battery type, and the charging time. Among them, the maximum actual current of the charging process can be used as the target supervision value and the test target value of the decision tree model.

[0036] In step S120 , training data is determined according to the charging behavior data of the vehicle and a clustering algorithm.

[0037] The training data is data used to train the decision tree model, and the training data may include input features and labels corresponding to the input features. In an embodiment of the present application, the training data is obtained by clustering the charging behavior data using a clustering algorithm.

[0038] In some embodiments, the charging behavior data includes location information, and the training data may be aggregated information of charging piles obtained by clustering the charging behavior data of vehicles according to a clustering algorithm and location information, where the aggregated information of the charging piles includes location tags.

[0039] Based on this, step S120 may include: clustering the charging behavior data of the vehicle according to the clustering algorithm and the location information to obtain the aggregate information of the charging piles, wherein the aggregate information of the charging piles includes the location tags; and using the aggregate information of the charging piles as training data.

[0040] In other embodiments, the training data may include the aggregate information of the charging pile and the map-attached information of the charging pile. For example, the training data may be the fusion information (or integrated information) of the aggregate information of the charging pile and the map-attached information of the charging pile. The map-attached information of the charging pile may be obtained based on the above-mentioned location tag, and the map-attached information of the charging pile is used to indicate various feature data related to the location of the charging pile. For example, the map-attached information of the charging pile includes, but is not limited to, any one or more of the following data: the type of charging pile, the density of commercial facilities within a preset distance of the charging pile, the density of residential areas within a preset distance of the charging pile, the construction time of the charging pile, the brand of the charging pile, and the popularity of the charging station, etc.

[0041] Based on this, step S120 may include: clustering the charging behavior data of the vehicle according to the clustering algorithm and the location information to obtain the aggregate information of the charging pile, the aggregate information of the charging pile includes the location tag; obtaining the map-attached information of the charging pile according to the location tag; integrating the aggregate information of the charging pile and the map-attached information of the charging pile into training data. By merging and integrating the aggregate information of the charging pile and the map-attached information of the charging pile into training data, the dimension of the data features of the training data can be increased, thereby further improving the output accuracy of the decision tree model.

[0042] The embodiment of the present application does not specifically limit the method for obtaining the map-attached information of the charging pile. For example, the map-attached information of the charging pile can be obtained from the map application programming interface (API) interface according to the location tag in the aggregated information of the charging pile. In this way, the map-attached information of the charging pile can be obtained by setting an API interface that can connect to the map application on the cloud platform or server that executes the training method. Not only is the implementation method simple, but the efficiency of obtaining the map-attached information of the charging pile is also high.

[0043] It should be noted that the location information in the charging behavior data is the charging location information of the vehicle, and the location tag is the location information of the charging pile obtained after clustering the charging location information. It can be seen that the clustering algorithm can aggregate the charging behavior of the vehicle into the charging behavior of the charging pile. Since the aggregation information of the charging pile is obtained through the clustering cluster algorithm, each charging pile will correspond to a cluster, and the number of the cluster is equivalent to the number of the charging pile. In some embodiments, the charging piles can also be renumbered according to the number of the cluster.

[0044] The embodiment of the present application does not specifically limit the type of clustering algorithm. For example, the clustering algorithm can be any one of the following algorithms: K-Means algorithm, hierarchical clustering algorithm, density-based noise-resistant clustering algorithm, etc. Preferably, the clustering algorithm can be a density-based noise-resistant clustering algorithm. For example, the clustering algorithm is a classic density-based clustering algorithm (density-based spatial clustering of applications with noise, DBSCAN). By selecting a density-based noise-resistant clustering algorithm, noise data and abnormal data can be better processed, so that the aggregated information of the charging piles is more accurate, which is conducive to improving the detection accuracy of the decision tree model. The density-based noise-resistant clustering algorithm can be specifically described in the following text.

[0045] In step S130, a decision tree model is trained according to the training data.

[0046] The initial trained decision tree model can be a manually set decision tree regression model. Alternatively, the initial trained decision tree model can be a DecisionTreeRegressor decision tree regression model called from sklearn. By directly calling an existing decision tree model as the initial trained decision tree model, the training efficiency of the training decision tree model can be improved.

[0047] In an embodiment of the present application, supervised machine learning can be performed on the decision tree model based on the training data. Based on this, it is also necessary to set test data and target test values, which are used to test the subsequently trained decision tree model. As mentioned above, the target test value can be the maximum actual current of the charging process in the charging behavior data of the vehicle. The method of obtaining the test data is consistent with the training method.

[0048] In step S140, the parameters of the decision tree model are updated according to the output of the decision tree model.

[0049] The parameters of the decision tree model may include, for example, one or more of the root node, internal node, branch (or path), and leaf node in the decision tree model. Updating the parameters of the decision tree model may include pruning, modifying internal nodes, modifying branches, modifying leaf nodes, etc.

[0050] The embodiment of the present application does not specifically limit the method of updating the parameters of the decision tree model according to the output of the decision tree model. As an example, the parameters of the decision tree model can be updated according to the difference between the output of the decision tree model and the target supervision value (for example, a loss function such as mean square error and cross entropy).

[0051] It should be noted that, in each training, the training data needs to be input into the decision tree model and the output of the decision tree model needs to be collected. After the output of the decision tree model is collected, it will be determined whether the difference between the output of the decision tree model and the target supervision value meets the requirements. If it does not meet the requirements (for example, the loss function does not converge), back propagation is performed to update the parameters of the decision tree model, and the decision tree model with updated parameters is continued to be trained. If it meets the requirements (for example, the loss function converges), it is not necessary to train again. The decision tree model at this time is the trained decision tree model. You can choose to save all the parameters of this decision tree model (for example, you can use grid search to obtain the parameters of the decision tree model at this time), and save the decision tree model as a binary file. The decision tree model is the decision tree model deployed on the cloud platform for estimating the prediction value described later.

[0052] In some embodiments, in order to further improve the prediction accuracy of the decision tree model, step S120 may include: preprocessing (or optimizing) the vehicle's charging behavior data to obtain processed vehicle charging behavior data; determining training data based on the processed vehicle charging behavior data and a clustering algorithm.

[0053] The embodiments of the present application do not specifically limit the preprocessing method. For example, preprocessing can be to integrate the charging behavior data of the vehicle, for example, to integrate the charging behavior data of the vehicle into a unified format to facilitate the learning of the decision tree model, and to improve the performance and generalization ability of the decision tree model. For another example, preprocessing can be to delete redundant information, erroneous information or abnormal information in the charging behavior data of the vehicle to improve the effectiveness of the training data, thereby improving the learning ability of the decision tree model and the final performance.

[0054] As an example, redundant information in the charging behavior data of the vehicle may be deleted according to the charging time in the charging behavior data of the vehicle. For example, the charging behavior data of the vehicle associated with the charging time that does not meet the requirements is deleted from the big data set formed by the charging behavior data of all vehicles as redundant information, wherein the charging time that does not meet the requirements may be a charging time that is less than a set threshold.

[0055] As another example, redundant information in the vehicle's charging behavior data can be deleted based on the maximum actual current of the charging process and the maximum limit current of the vehicle's charging process in the vehicle's charging behavior data. For example, the vehicle's charging behavior data associated with the maximum actual current of the charging process that does not meet the requirements is deleted from the big data set formed by all the vehicle's charging behavior data as redundant information, wherein the maximum actual current of the charging process that does not meet the requirements is the value of the maximum actual current of the charging process>(the maximum limit current of the vehicle's charging process*90%).

[0056] The above describes in detail the training method of the decision tree model provided in the embodiment of the present application. Figure 2 The charging indicator evaluation method provided in the embodiment of the present application is received in detail. Figure 2 As shown, the charging indicator evaluation method provided in the embodiment of the present application includes steps S210 to S240.

[0057] In step S210, the vehicle sends a charging signal to the cloud platform.

[0058] The vehicle is equipped with a BMS, and the vehicle can send a charging signal to the cloud platform through the BMS when preparing for charging or starting charging. The charging signal is used to indicate that the vehicle is about to charge or starts charging. The charging signal may include charging data. The charging data includes but is not limited to any one or more of the following information: location information (e.g., latitude information and longitude information), vehicle information (e.g., vehicle brand, vehicle model information (or vehicle model), etc.), battery type, charging time, remaining battery power, etc.

[0059] The cloud platform may also be referred to as a vehicle service platform, and the cloud platform may refer to a cloud-based management platform maintained by a vehicle manufacturer. For example, the cloud platform may be a telematics service provider (TSP) cloud platform.

[0060] In step S220, the cloud platform inputs the charging data in the charging signal into the decision tree model to obtain a predicted value output by the decision tree model.

[0061] In the embodiment of the present application, a decision tree model (or decision tree regression model) is deployed in the cloud platform. The cloud platform can collect the charging behavior data of the vehicle reported by each vehicle. The decision tree model deployed in the cloud platform is a machine learning model obtained after training based on a large amount of vehicle behavior data. For example, the decision tree model deployed in the cloud platform can be a decision tree model obtained according to the training method described above.

[0062] The embodiment of the present application does not specifically limit the type of decision tree model. For example, the decision tree model can be a DecisionTreeRegressor decision tree regression model or a CART decision tree regression model in sklearn.

[0063] In step S230, the cloud platform sends the predicted value to the vehicle.

[0064] The embodiments of the present application do not specifically limit the interaction method between the cloud platform and the vehicle.

[0065] For example, the cloud platform can interact with the vehicle through a wireless network. For example, the wireless network is any one of a 4G communication network, a 5G communication network, and a satellite communication network. Based on this, the cloud platform can send the predicted value to the vehicle through a downlink channel, and the vehicle can send the charging signal to the cloud platform through an uplink channel.

[0066] For another example, the cloud platform can interact with the vehicle through an intermediate device. The intermediate device is a device that can interact with the cloud platform and the vehicle, for example, the intermediate device is a terminal device (such as a mobile phone). For another example, the intermediate device is a network device (such as a base station).

[0067] In step S240 , the vehicle evaluates the charging index according to the predicted value.

[0068] The charging index may indicate relevant information about the charging of the vehicle, which may be used to monitor battery safety and improve user experience. The present application embodiment does not specifically limit the charging index. For example, the charging index includes but is not limited to any one or more of the following: remaining charging time, state of charge (SOC), battery temperature, etc.

[0069] In the charging index evaluation method provided in the embodiment of the present application, the vehicle can interact with a cloud platform deployed with a decision tree model, wherein the decision tree model can automatically predict the output current capacity value of the charging pile for charging the vehicle based on the charging data in the charging signal sent by the vehicle, and the vehicle can evaluate the charging index based on the predicted value output by the decision tree model. In this way, the problem of errors in the output current capacity value of the charging pile obtained by the vehicle from the charging pile can be avoided, thereby improving the accuracy of the charging index evaluated by the vehicle, improving the charging experience and effectively ensuring charging safety.

[0070] The embodiments of the present application do not specifically limit the algorithm of the decision tree model, which is related to the training process and training data.

[0071] In some implementations, the decision tree model is trained based on the training data described above, and the training data is obtained based on a large amount of vehicle charging behavior data collected by the cloud platform and combined with a clustering algorithm. In view of this, the charging data may include location information, and step S220 may include: inputting the charging data into the decision tree model so that the decision tree model performs the following operations: determining the map-attached information of the charging pile according to the location information, the map-attached information of the charging pile includes any one or more of the following data: the type of charging pile, the density of commercial facilities within a preset distance of the charging pile, the density of residential areas within a preset distance of the charging pile, and the construction time of the charging pile; outputting a predicted value based on the map-attached information of the charging pile and the charging data.

[0072] As an example, the map-attached information of the charging pile includes the type of the charging pile and the density of residential areas within a preset distance of the charging pile, and the charging data also includes vehicle type information. Outputting the predicted value based on the map-attached information of the charging pile and the charging data may include: determining a first target output path based on the type of the charging pile, the first target output path being one of a plurality of first output paths, and the plurality of first output paths corresponding one-to-one to the groupings of the types of the plurality of charging piles; in response to the first target output path indicating that the predicted value needs to be output, outputting the predicted value from the first target output path; in response to the first target output path indicating that the second target output path needs to be determined, the predicted value is output based on the density of residential areas and the first target output path. The method comprises the following steps: determining a second target output path according to the vehicle model information and the second target output path, wherein the second target output path is one of a plurality of second output paths, and the plurality of second output paths correspond one-to-one to the groups of the plurality of residential area densities; in response to the second target output path indicating that a predicted value needs to be output, outputting the predicted value from the second target path; in response to the second target output path indicating that a third target output path needs to be determined, determining the third target output path according to the vehicle model information and the second target output path, wherein the third target output path is one of a plurality of third output paths, and the plurality of third output paths correspond one-to-one to the groups of the plurality of vehicle model information; in response to the third target output path indicating that a predicted value needs to be output, outputting the predicted value from the third target output path.

[0073] For example, Figure 3 As shown, there are 3 first output paths in total, and the 3 first output paths correspond to the types A, B and C of the charging piles respectively. If the first target output path is a path corresponding to the type C of the charging pile, the first target output path indicates a direct leaf node, that is, the predicted value 400A needs to be output. If the first target output path is a path corresponding to the type A or B of the charging pile, the first target output path indicates that the second target output path needs to be determined. The second output paths include 4, including 2 associated with the first output path (associated A) and corresponding to population densities 10 and 20, and 2 associated with the first output path (associated B) and corresponding to population densities 4 and 20. There are 2 third output paths, including two associated with the second output path (associated A and associated with population density 20) and corresponding to vehicle models M and N. In order to facilitate the understanding of the above training process and the output process of the predicted value, the following is combined with Figure 4-Figure 6 The training and use methods of the decision tree model in the embodiment of the present application are explained.

[0074] like Figure 4 As shown, the training and use method of the decision tree model provided in the embodiment of the present application includes steps S410-S460.

[0075] In step S410, the charging behavior data of the vehicle is collected through big data.

[0076] The vehicle's charging behavior data mainly includes the charging location longitude information, the charging location latitude information, the maximum actual current during the charging process, the maximum limited current during the vehicle charging process, vehicle model, battery type, charging time and other data characteristics.

[0077] In step S420: optimizing the charging behavior data of the vehicle.

[0078] The method of optimizing the charging behavior data of the vehicle includes data cleaning (deletion) and / or integration of features of the charging behavior data of the vehicle.

[0079] As an implementation method, the cleaning may be: the set formed by the charging behavior data of the vehicle in step S410 is recorded as: U all , select the data set formed by the maximum actual current of the charging process>(the maximum limit current of the vehicle charging process*90%) and record it as: X lim , through U all –X lim The obtained data set R is used as the optimized data.

[0080] In step S430: the charging pile information is aggregated using the DBSCAN algorithm.

[0081] Step S430 may specifically be to aggregate the charging pile information using the DBSCAN algorithm and the longitude and latitude information in the optimization data.

[0082] like Figure 5 As shown, the implementation of the DBSCAN algorithm includes the following steps.

[0083] In step S510: select any point p from the set R as the center point.

[0084] In step S520: make a judgment based on t(∈, MinPts).

[0085] t(∈, MinPts) is the key parameter of the DBSCAN algorithm, which is used to describe the compactness of sample distribution in the neighborhood, where MinPts is the minimum number of cluster points and ∈ is the cluster radius.

[0086] Step S520 specifically includes the following contents.

[0087] Step S521, taking point p as the center point and t(∈, MinPts) as the key parameter, determine whether the number of points in the radius area ∈ exceeds MinPts.

[0088] If it is greater than or equal to MinPts, then execute step S522: determine that point p is a core point, put it into the core point set M, and put points other than point p into the adjacent point set N.

[0089] If the result of the judgment is less than MinPts, then execute step S523: judge whether the point is in the adjacent point set N. If so, execute step S524: put point P into the edge point set Y; if not, execute step S525: put point p into the noise point set U.

[0090] Step S526, take a point p as the center from the set of adjacent points N without replacement, and repeat steps S521-S525.

[0091] Step S530: until the neighboring point set N is empty, transfer all points in the core point set M and the edge set to the cluster point set X. i , which is the i-th cluster point set.

[0092] Step S540: subtract the noise point set U from the set of all points R, and then subtract the currently calculated X0, X1, ..., X i After the cluster points are gathered, iterative calculations are started from S510 until the set R is empty.

[0093] Through the above method, the charging behavior of the vehicle is synthesized into the charging behavior of the charging station, and the specific charging piles are numbered in the data. For example, the number of the cluster point set finally obtained is the number of the charging pile.

[0094] For ease of understanding, Figure 6 A schematic diagram of the calibration of a cluster point set obtained using the DBSCAN algorithm is shown.

[0095] Step S440: Add data features of the charging pile dimension through the map API.

[0096] S440 specifically obtains map-attached information of the charging piles through a commercial map (such as Amap, Baidu Map) API according to the aggregated charging pile center location. The map-attached information of the charging piles is the data feature of the added charging pile dimension.

[0097] The map-attached information of the charging pile includes, but is not limited to: the brand type of the charging pile, the density of nearby commercial facilities, the density of residential areas, the time of station establishment, etc. The data features of the charging pile dimension can be increased through step S440.

[0098] Step S450: training the decision tree regression model to obtain a trained decision tree regression model.

[0099] Step S450, for example, includes summarizing and organizing the map-attached information of the cluster point set and the charging pile to train the DecisionTreeRegressor decision tree regression model in sklearn, and storing the trained decision tree regression model after obtaining the optimal model parameters. Specifically, it may include: summarizing and organizing the map-attached information of the cluster point set and the charging pile into training data, test data, and target supervision values ​​and target test values. The target supervision value and the target test value are, for example, the maximum output current capacity value during the charging process. Part of the other data features is training data, and the other part is test data. Supervised machine learning model training is performed through the training data, and grid search is used to obtain the optimal parameter configuration items of the optimal DecisionTreeRegressor decision tree regression model. After the training is completed, the trained model is trained based on the test data. If the test is completed, the trained model is saved as an importable binary file.

[0100] The specific logic diagram of the decision tree model can be as follows Figure 3 shown.

[0101] Step S460: Predict the maximum output current capacity value of the charging pile where the vehicle is to be charged through the trained decision tree model.

[0102] The existing decision tree is based on the decision tree model trained in step S450. The model can be deployed on a cloud platform so that the cloud platform can predict the maximum output current capacity of the charging pile for the vehicle to be charged and send it to the vehicle so that the vehicle can calculate key indicators such as the remaining charging time and SOC during the actual charging process.

[0103] The embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed, the aforementioned method steps are implemented.

[0104] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0105] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0106] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0107] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0108] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part 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, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that can be read by a computer or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital video disc (DVD)), or a semiconductor medium (eg, a solid state disk (SSD)).

[0109] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A charging index evaluation method, characterized in that: The method is applied to a cloud platform, in which a decision tree model is deployed, and the method includes: receiving a charging signal sent by a vehicle, wherein the charging signal includes charging data; Inputting the charging data into the decision tree model to obtain a predicted value output by the decision tree model, wherein the predicted value is used to characterize an output current capacity value of a charging pile supplying power to the vehicle; The predicted value is sent to the vehicle, so that the vehicle evaluates the charging index according to the predicted value.

2. The method according to claim 1, characterized in that The charging data includes location information, and the step of inputting the charging data into the decision tree model to obtain a predicted value output by the decision tree model includes: The charging data is input into the decision tree model so that the decision tree model performs the following operations: Determining map-attached information of the charging pile according to the location information, the map-attached information of the charging pile including any one or more of the following data: the type of the charging pile, the density of commercial facilities within a preset distance of the charging pile, the density of residential areas within a preset distance of the charging pile, and the establishment time of the charging pile; The predicted value is output according to the map-attached information of the charging pile and the charging data.

3. The method according to claim 2, characterized in that The map-attached information of the charging pile includes the type of the charging pile and the density of residential areas within a preset distance of the charging pile, the charging data includes vehicle type information, and outputting the predicted value according to the map-attached information of the charging pile and the charging data includes: Determining a first target output path according to the type of the charging pile, the first target output path being one of a plurality of first output paths, and the plurality of first output paths corresponding one-to-one to the groups of the types of the plurality of charging piles; In response to the first target output path indicating that the predicted value needs to be output, outputting the predicted value from the first target output path; In response to the first target output path indicating that a second target output path needs to be determined, the second target output path is determined according to the residential area density and the first target output path, the second target output path being one of a plurality of second output paths, and the plurality of second output paths corresponding one-to-one to a plurality of groups of the residential area densities; In response to the second target output path indicating that the predicted value needs to be output, outputting the predicted value from the second target path; In response to the second target output path indicating that a third target output path needs to be determined, the third target output path is determined according to the vehicle type information and the second target output path, the third target output path being one of a plurality of third output paths, and the plurality of third output paths corresponding one-to-one to the groups of the plurality of vehicle type information; In response to the third target output path needing to output the predicted value, the predicted value is output from the third target output path.

4. The method according to any one of claims 1 to 3, characterized in that: The decision tree model is obtained by training based on training data, and the training data is obtained based on vehicle charging behavior data and a clustering algorithm.

5. A charging index evaluation method, characterized in that: The method is applied to a vehicle, and comprises: Sending a charging signal to a cloud platform, wherein the charging signal includes charging data; Receiving a prediction value sent by the cloud platform, where the prediction value is used to characterize an output current capability value of a charging pile that supplies power to the vehicle and the prediction value is obtained by a decision tree model deployed in the cloud platform based on the charging data; The charging indicator is evaluated according to the predicted value.

6. A training method for a decision tree model, characterized in that: The method is used to train a decision tree model as described in any one of claims 1 to 4, and the method comprises: Obtain vehicle charging behavior data; Determining training data based on the charging behavior data of the vehicle and a clustering algorithm; Training the decision tree model according to the training data; The parameters of the decision tree model are updated according to the output of the decision tree model.

7. The method according to claim 6, characterized in that The charging behavior data of the vehicle includes location information, and the determining of training data according to the charging behavior data of the vehicle and a clustering algorithm includes: Clustering the charging behavior data of the vehicle according to a clustering algorithm and the location information to obtain aggregated information of the charging pile, wherein the aggregated information of the charging pile includes a location tag; Acquire map-related information of the charging pile according to the location tag; The aggregate information of the charging pile and the map-attached information of the charging pile are integrated into the training data.

8. The method according to claim 6, characterized in that The determining of training data according to the charging behavior data of the vehicle and the clustering algorithm comprises: Preprocessing the charging behavior data of the vehicle to obtain processed charging behavior data of the vehicle; The training data is determined according to the processed charging behavior data of the vehicle and the clustering algorithm.

9. The method according to claim 7, characterized in that: The acquiring of the map-attached information of the charging pile according to the location tag includes: According to the location tag, map-attached information of the charging pile is obtained through a map application programming interface (API) interface, and the map-attached information of the charging pile includes any one or more of the following data: the type of the charging pile, the density of commercial facilities within a preset distance of the charging pile, the density of residential areas within a preset distance of the charging pile, and the establishment time of the charging pile.

10. A computer-readable storage medium, characterized in that: The computer storage medium stores a computer program, and when the computer program is executed, the method according to any one of claims 1 to 4 or claim 5 or claims 6 to 9 is implemented.

Images