Charging station idleness prediction method and device

Abstract

The present invention provides a method and device for predicting the idleness of a charging station. The method comprises: obtaining a first label corresponding to each target charging pile contained in the target charging station according to an identifier of the target charging station; determining an idleness prediction result of the target charging station according to the first label corresponding to each target charging pile contained in the target charging station and a prediction model obtained by pre-training; since the prediction model takes into account the influence of multiple factors on the idleness, such as the scale of the charging station, the evaluation score of the charging pile, the charging price, the charging mode, and the parking fee information, during training, the idleness prediction result of the charging pile obtained by the prediction model is more accurate, thereby making the idleness prediction result of the charging station more accurate.

Description

Technical Field

[0001] The present invention relates to the field of charging technology services, and in particular to a method and device for predicting the idleness of a charging station. Background Art

[0002] New energy vehicles refer to vehicles that use unconventional vehicle fuels as a power source (or use conventional vehicle fuels but adopt new on-board power devices), integrate advanced technologies in vehicle power control and drive, and form vehicles with advanced technical principles, new technologies, and new structures. New energy vehicles include pure electric vehicles, extended-range electric vehicles, hybrid vehicles, fuel cell electric vehicles, and hydrogen engine vehicles. Among them, pure electric vehicles are more common in daily life. For owners of pure electric vehicles, when the vehicle is low on power, they need to find a charging station to charge the vehicle. If the owner can understand the probability that the charging station is currently or in the future, it can help the owner find an idle charging pile in the shortest time, thereby solving the pain point of the owner looking for an idle charging pile and improving the user experience.

[0003] In the prior art, the proportion of time that each charging pile in the charging station is idle in the previous period is obtained, and then the average value is calculated and used as the idleness of the charging station on the next day. However, the idleness of the charging station is affected by many factors, and the idleness obtained by the above method is not accurate. Summary of the invention

[0004] The present invention provides a charging station idleness prediction method and device, which are used to improve the accuracy of charging station idleness prediction results.

[0005] In a first aspect, the present invention provides a method for predicting the idleness of a charging station, comprising: obtaining, according to an identification of a target charging station, a first label corresponding to each target charging pile contained in the target charging station, the first label comprising at least one of the following labels: the scale of the target charging station, the evaluation score of the target charging pile, the charging price of the target charging pile, the charging mode of the target charging pile, the parking fee information of the parking space corresponding to the target charging pile, the predicted time period, the week attribute corresponding to the predicted date, or the holiday attribute corresponding to the predicted date; determining a prediction result of the idleness of the target charging station according to the first label corresponding to each target charging pile contained in the target charging station and a pre-trained prediction model; and outputting the prediction result of the idleness.

[0006] In a second aspect, the present invention provides a charging station idleness prediction device, comprising: a receiving module for receiving an idleness query request triggered by a user, the idleness query request comprising: an identifier of a target charging station requested to be queried by the user; an acquisition module for acquiring, based on the identifier of the target charging station, a first tag corresponding to each target charging pile contained in the target charging station, the first tag comprising at least one of the following tags: the scale of the target charging station, the evaluation score of the target charging pile, the charging price of the target charging pile, the charging mode of the target charging pile, the parking fee information of the parking space corresponding to the target charging pile, the predicted time period, the week attribute corresponding to the predicted date, or the holiday attribute corresponding to the predicted date; a determination module for determining an idleness prediction result of the target charging station based on the first tag corresponding to each target charging pile contained in the target charging station and a pre-trained prediction model; and an output module for outputting the idleness prediction result.

[0007] In a third aspect, the present invention provides a charging pile platform on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned charging station idleness prediction method is implemented.

[0008] In a fourth aspect, the present invention provides a terminal device, comprising the above-mentioned charging pile platform.

[0009] The present invention provides a prediction and device for the idleness of charging piles. On the one hand, the historical idleness of all existing charging piles is obtained, and on the other hand, the labels of all charging piles are obtained, and the labels include: the scale of the charging station, the ratings of the charging piles by previous users, the charging price of the charging piles and / or the charging mode of the charging piles and other information. The above-collected data is used as sample data, and a prediction model is obtained through training. When a user triggers a request to view the idleness of a charging station, the label of each charging pile contained in the charging station is input into the prediction model, and the idleness prediction results of all charging piles can be obtained. Then, the idleness prediction results of all charging piles are averaged to obtain the idleness prediction results of the charging station. Since the influence of various factors on the idleness is taken into account during model training, the idleness prediction results of the charging piles obtained by the prediction model are more accurate, and thus the idleness prediction results of the charging station are also more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 Provide an optional system architecture diagram for the present invention;

[0011] Figure 2 A signaling flow chart of the first embodiment of the charging station idleness prediction method provided by the present invention;

[0012] Figure 3 User interface provided by the present invention Figure 1 ;

[0013] Figure 4 User interface provided by the present invention Figure 2 ;

[0014] Figure 5 A schematic diagram of a flow chart of a second embodiment of a charging station idleness prediction method provided by the present invention;

[0015] Figure 6 A schematic diagram of the structure of the charging station idleness prediction device provided by the present invention;

[0016] Figure 7 A schematic diagram of the hardware structure of the server provided by the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] For electric car owners, if they can understand the idleness of a charging station in a certain period of time in the future, it can provide a reference for finding a charging station for travel, and then quickly find an idle charging station, saving time in finding a charging station. In the prior art, in order to obtain the idleness of a charging station, the time proportion of each charging station contained in the charging station in the previous period of time is first obtained, and the time proportion is used as the idleness of the corresponding charging station on the second day. The average of the idleness of all charging piles is calculated, and the average value is used as the idleness of the charging station on the second day. However, the idleness of the charging pile is affected by many factors, and the accuracy of the idleness obtained by the above method is not high.

[0019] Figure 1 An optional system architecture diagram is provided for the present invention. Figure 1 The system shown includes: a terminal device and a server. The terminal device and the server are wirelessly connected. The terminal device can be a mobile phone, a tablet computer, a smart watch, a television, or other electronic devices with a display screen. As an example but not a limitation, the terminal device can also be a wearable device. Figure 1 The terminal device is represented by a mobile phone, and the server can be a physical server or a cloud server.

[0020] Optionally, corresponding application software can be installed on the terminal device. The user can trigger an idleness query request for a charging station through the application software. The terminal device can send the request to the server. The server can predict the idleness of the charging station by executing the prediction method provided by the present invention. After obtaining the prediction result, the server can further output the prediction result to the terminal device, and the terminal device will render and display the prediction result.

[0021] The idleness of the charging piles in the charging station is affected by many factors, such as: the scale of the charging station, the previous user ratings of the charging piles, the charging price of the charging piles, and the charging mode of the charging piles. In the prior art, the proportion of time that the charging piles are idle in the previous period is directly used as the probability that the charging piles will be idle the next day, without considering the influence of the above factors on the idle probability of the charging piles, resulting in low accuracy of the idleness obtained by the prior art. The present invention takes this as a starting point, on the one hand, obtains the historical idleness of all existing charging piles, and on the other hand, obtains the labels of all charging piles, which include: the scale of the charging station, the previous user ratings of the charging piles, the charging price of the charging piles and / or the charging mode of the charging piles. The above collected data is used as sample data, and a prediction model is obtained through training. When a user triggers a request to view the idleness of a charging station, the label of each charging pile contained in the charging station is input into the prediction model, and the idleness prediction results of all the charging piles can be obtained. Then, the idleness prediction results of all the charging piles are averaged to obtain the idleness prediction results of the charging station. Since the impact of various factors on the idleness is taken into account during model training and an appropriate algorithm is selected, the idleness prediction results of charging piles obtained through the prediction model are more accurate, which in turn makes the idleness prediction results of charging stations more accurate.

[0022] The charging station idleness prediction method provided by the present invention is described in detail below in conjunction with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present invention will be described below in conjunction with the accompanying drawings.

[0023] based on Figure 1 The system architecture shown, Figure 2 This is a signaling flow chart of the first embodiment of the charging station idleness prediction method provided by the present invention. Figure 2 As shown, the charging station idleness prediction method provided in this embodiment includes:

[0024] S201. The terminal device receives an idleness query request triggered by a user.

[0025] In a possible implementation, an application software for querying the idleness of a charging station can be installed on the terminal device, see Figure 3As shown, the application can display a map with the distribution of charging stations. Figure 3 There are 4 charging stations distributed on the map, which are respectively identified as ID1, ID2, ID3 and ID4. The user can trigger an idleness query request for the charging station by clicking on the icon of the charging station distributed on the map.

[0026] It should be noted that the idleness query request triggered by the user includes the identifier of the target charging station requested by the user, for example: Figure 3 The operation shown triggers an idleness query request, and the idleness query request includes an identifier ID1.

[0027] S202: The terminal device sends an idleness query request to the server.

[0028] S203: The server obtains a first tag corresponding to each target charging pile included in the target charging station according to the identifier of the target charging station.

[0029] Among them, the first tag corresponding to the target charging pile includes at least one of the following tags: the scale of the charging station where the target charging pile is located, the evaluation score of the target charging pile, the charging price of the target charging pile, the charging mode of the target charging pile, the parking fee information of the parking space corresponding to the target charging pile, the predicted time period, the week attribute corresponding to the predicted date, or the holiday attribute corresponding to the predicted date.

[0030] In a possible implementation, the server stores tags corresponding to the charging piles included in all charging stations that have been put into use. When the server receives an idleness query request, it searches all the stored charging stations according to the identifier of the target charging station in the idleness query request, thereby obtaining tags such as the evaluation score of each target charging pile included in the target charging station, the charging price of the target charging pile, the charging mode of the target charging pile, and the parking fee information of the parking space corresponding to the target charging pile. As for the predicted time period, the week attribute corresponding to the predicted date, or the holiday attribute corresponding to the predicted date, the server can obtain it according to the default settings, for example, the predicted time period can be set to the time period corresponding to each hour in the next 24 hours, for example, the date when the server receives the idleness query request is 2020-5-1, and the time is 12:00, then the server can determine that the predicted time period is 12:00-13:00, 13:00-14:00, ... from 2020-5-1, 12:00, a total of 24 hours. The week attribute corresponding to the predicted date can be set to the week attribute corresponding to the date when the server receives the availability query request. Continuing with the above example, the date when the server receives the availability query request is 2020-5-1, and 2020-5-1 is Friday, then the server can determine that the week attribute corresponding to the predicted date is "Friday". The holiday attribute corresponding to the predicted date can be set to the holiday attribute corresponding to the date when the server receives the availability query request. Continuing with the above example, the date when the server receives the availability query request is 2020-5-1, and 2020-5-1 is a holiday, then the server can determine that the holiday attribute corresponding to the predicted date is "yes".

[0031] In another possible implementation, after the server receives the idleness query request sent by the terminal device, it returns a tag input instruction to the terminal device. After the terminal device receives the tag input instruction, it prompts the user to fill in the first tag corresponding to the target charging pile in the form of a dialog box, and then feeds back the first tag corresponding to the target charging pile to the server.

[0032] S204: Determine an idleness prediction result of the target charging station according to a first label corresponding to each target charging pile included in the target charging station and a pre-trained prediction model.

[0033] First, according to the first label corresponding to each target charging pile contained in the target charging station and the prediction model, the idleness prediction results of all target charging piles are obtained. Then, the idleness prediction results of all target charging piles are averaged to obtain the idleness prediction result of the target charging station.

[0034] In one possible implementation, after receiving the idleness query request, the server first obtains the first tag corresponding to each target charging pile contained in the target charging station requested by the user through S203, and then inputs the first tag corresponding to the target charging pile into a pre-trained prediction model to obtain the idleness prediction results of all target charging piles.

[0035] It should be noted that the dimension of the first label in the input prediction model needs to be consistent with the label dimension used when training the prediction model. For example, when training the prediction model, the labels used are: the scale of the charging station, the evaluation score of the charging pile, the charging price of the charging pile, the charging mode of the charging pile, the time period and the weekly attribute. Then, when the idleness of the target charging pile is predicted by the prediction model, the label input to the prediction model needs to be: the scale of the charging station where the target charging pile is located, the evaluation score of the target charging pile, the charging price of the target charging pile, the charging mode of the target charging pile, the predicted time period and the weekly attribute corresponding to the predicted date. Alternatively, there are multiple pre-trained prediction models, and each prediction model uses different dimensions when training. After the first label corresponding to the target charging pile is obtained, based on the dimension of the first label, the prediction model applicable to the first label is selected for idleness prediction.

[0036] S205. The server outputs the idleness prediction result to the terminal device.

[0037] S206. The terminal device displays the idleness prediction result of the target charging station.

[0038] Combination Figure 3 For example, as described above, the server can set the prediction time period to the time period corresponding to each hour in the next 24 hours. Figure 3 After triggering the idleness query request of ID1, the server obtains the first tag corresponding to each hour of each target charging pile contained in ID1 within the next 24 hours, and then inputs the first tag corresponding to each hour into the prediction model to obtain the idleness prediction result of each target charging pile within the next 24 hours. The idleness prediction results of all target charging piles are averaged to obtain the idleness prediction result of the charging station within the next 24 hours. The server can further process the difference of the idleness prediction results of the 24 hours. After the server sends the prediction result to the terminal device, the terminal device renders and displays it to obtain Figure 4 The results are shown, and users can use them as a reference for finding charging piles when traveling.

[0039] It should be noted that: S203-S206 describes the solution of calculating the idleness of the corresponding charging station in real time and feeding it back to the user after the user triggers the idleness query request. Real-time calculation requires relatively high computing power. In the case of limited computing power, S203-S206 can also be replaced by the following solution:

[0040] In one possible implementation, the server regularly predicts the idleness of each charging station, for example, once a day or once an hour, and then stores the idleness prediction results. After the user triggers a query request for querying the idleness of the target charging station, the idleness prediction results can be directly retrieved from the stored idleness prediction results based on the query request, thereby reducing the requirements for computing power.

[0041] In another possible implementation, when a user triggers an idleness query request, the server calculates and stores the idleness of the corresponding charging station. When another user requests to query the idleness of the same charging station, the stored idleness is directly fed back to the user. This can also reduce the requirements for computing power.

[0042] It should be noted that: This embodiment is based on Figure 1 The method provided by the present invention is explained by taking the shown architecture as an example. The various steps in this embodiment may also be completely executed by the terminal device, that is, S203-S206 using the server as the execution subject is only an example. S203-S206 may also be executed by the terminal device, and this embodiment is not limited here.

[0043] The charging station idleness prediction method provided by the present invention obtains the idleness prediction result of the target charging pile through a prediction model. Since the prediction model takes into account the influence of multiple factors on the idleness, such as the scale of the charging station, the charging pile evaluation score, the charging price, the charging mode and the parking fee information, during training, the charging pile idleness prediction result obtained by the prediction model is more accurate, and the charging station idleness prediction obtained based on the charging pile idleness prediction result is also more accurate.

[0044] The following is a detailed description of the process by which the server determines the idleness prediction result of the target charging station. Figure 5 This is a flow chart of the second embodiment of the charging station idleness prediction method provided by the present invention. The charging station idleness prediction method provided by this embodiment can be Figure 1 to execute on the server in the Figure 5 As shown, the charging station idleness prediction method provided in this embodiment includes:

[0045] S501-S506 is the model training process:

[0046] S501, obtain N sample data.

[0047] Specifically, the historical idleness of any charging pile in each hour within a preset time range can be obtained; then the corresponding second tag of the charging pile in each hour within the preset time range is obtained, and the second tag includes at least one of the following tags: the scale of the charging station where the charging pile is located, the evaluation score of the charging pile, the charging price of the charging pile, the charging mode of the charging pile, the parking fee information of the parking space corresponding to the charging pile, the time period corresponding to the historical idleness, the weekly attribute corresponding to the historical idleness, or the holiday attribute corresponding to the historical idleness. The historical idleness of the charging pile in any hour and the corresponding second tag in any hour can be determined as a sample data.

[0048] Specifically, the above method of obtaining the historical idleness of the charging pile every hour may include: obtaining status information of the charging pile, the status information including: idle, off-grid, faulty and / or plug-in charging status, the plug-in charging status including plug-in not charging or plug-in charging; determining the historical idleness based on the status information every hour.

[0049] The following is an example of how to obtain historical idleness:

[0050] Assume that the status information of charging pile A from 12:00 to 13:00 in the past day (April 1, 2020) includes: idle and faulty, among which 12:00-12:30 is idle and 12:30-13:00 is faulty, then the historical idleness of charging pile A corresponding to 12:00-13:00 on April 1, 2020 is 50%.

[0051] The following is an example of the second label:

[0052] In the past, each hour corresponds to a historical idleness and a second label. Taking the above time period of 12:00-13:00 on April 1, 2020 as an example, assuming that the scale of the charging station where charging pile A is located is 50, the evaluation score of charging pile A is 4, the charging price of charging pile A is 20 yuan, the charging mode of charging pile A is fast charging, the parking fee of charging pile A is 0 yuan, April 1, 2020 is Friday, and April 1, 2020 is a holiday. Then the second label of charging pile A includes: scale (50), evaluation score (4), charging price (20), charging mode (fast charging), parking fee (0), week attribute (Friday), holiday attribute (holiday), time period (12:00-13:00).

[0053] Then, as shown in Table 1, the historical idleness (50%) and the second label (scale (50), evaluation score (4), charging price (20), charging mode (fast charging), parking fee (0), week attribute (Friday), holiday attribute (holiday), time period (12:00-13:00)) corresponding to the time period of 12:00-13:00 on April 1, 2020 can be used as a sample data.

[0054] Table 1

[0055]

[0056] It should be noted that the older the sample data is, the lower the accuracy of the prediction model trained by the sample data in predicting future idleness. Therefore, when training the prediction model, you can choose sample data with a more recent time, for example, you can set the preset time range to the last 3 months.

[0057] In addition, the second tag may also include: the area where the charging pile is located, the operator corresponding to the charging pile, the facilities around the charging pile, the frequency of oil trucks occupying the place, etc. The present invention takes the second tags of 8 dimensions shown in Table 1 as an example to illustrate the solution of the present invention.

[0058] S502: Use a hash splitting algorithm to divide N sample data into a training group and a verification group.

[0059] In a possible implementation, the label of each sample data in the N sample data can be standardized first, and then the standardized N sample data can be divided into a training group and a validation group. Each label is standardized to 0 or 1. For example, when the scale is greater than 15, it can be standardized to 1, and when it is less than or equal to 15, it can be standardized to 0; when the evaluation score is greater than 3, it can be standardized to 1, and when it is less than or equal to 3, it can be standardized to 0; when the charging price is greater than 10, it can be standardized to 1, and when it is less than or equal to 10, it can be standardized to 0; when the charging mode is fast charging, it can be standardized to 1, and when the charging mode is slow charging, it can be standardized to 0; when the parking fee is greater than 0, it can be standardized to 1, and when it is equal to 0, it can be standardized to 0; when the week attribute is a weekend, it can be standardized to 1, and when it is not a weekend, it can be standardized to 0; when the holiday attribute is a holiday, it can be standardized to 1, and when the holiday attribute is a non-holiday, it can be standardized to 0; when the time period is during the day, it can be standardized to 1, and when the time period is at night, it can be standardized to 0. According to this rule, the second label of the samples shown in Table 1 can be standardized as shown in Table 2:

[0060] Table 2

[0061]

[0062] In one possible implementation, when dividing the training group and the verification group, the charging pile IDs corresponding to the N sample data are first converted into hash values ​​through the MD5 Message-Digest Algorithm (MD5 for short), and then the obtained hash values ​​are mapped to the interval of 0-100000 through a hash operation, and the sample data corresponding to the last digit 00-59 are classified as the training group, and the sample data corresponding to the last digit 60-99 are classified as the verification group.

[0063] S503: Perform uniformity check on the training group and the validation group to obtain uniformity check results.

[0064] S504: Adjust the training group and the verification group respectively according to the uniformity check result to obtain a training sample set and a verification sample set.

[0065] Specifically, the purpose of adjusting the training group and the validation group according to the uniformity check result is to ensure that the proportion of any label with the same value is close in the two groups. For example, the proportion of samples with a scale value of 1 in the training group is 80%. The goal of the above adjustment is to ensure that the proportion of samples with a scale value of 1 in the validation group is 75%-85%.

[0066] It should be noted that the above-mentioned ratio range can be flexibly set according to actual conditions. For example, the proportion of the number of samples with a scale value of 1 in the training group is 80%, and the proportion of the number of samples with a scale value of 1 in the verification group can also be adjusted to 78%-82%, or adjusted to the same as the training group (i.e., 80%). The present invention is not limited to this.

[0067] S505: After initializing the weights corresponding to the labels in the neural network model, the neural network model is trained using the training sample set to obtain an optimized neural network model.

[0068] In one possible implementation, when initializing the weights, the weights corresponding to the labels of the above eight dimensions in the neural network can be set to equal values, that is, the weights of each dimension are set to 0.125.

[0069] In a possible implementation, when training a neural network model using a training sample set, the neural network model can be set to a 4-layer structure, including: an input layer, two hidden layers, and an output layer, the activation function of the hidden layer is a linear rectifier function ReLU, and the activation function of the output layer is a linear function Liner. The algorithm used in the above 4-layer neural network model can select a feedforward neural network algorithm, and select a variance cost function as the cost function of the back propagation algorithm.

[0070] S506: Use a verification sample set to verify the optimized neural network model. If the proportion of successfully verified sample data in the verification sample set exceeds a preset value, the optimized neural network model is determined as the prediction model.

[0071] In a possible implementation, after obtaining the optimized neural network model in S504, for each sample data in the verification sample set, each sample data contains a second label and a real idleness, the second label in the sample is input into the above neural network model, and after obtaining the corresponding predicted value, the predicted value is compared with the real idleness in the sample data. If the error is within 5%, it is considered that the prediction is successful. The above operation is performed on each sample data in the verification sample set. If more than 90% of the sample data in the verification sample set are predicted successfully, the optimized neural network model is determined as the prediction model. If the proportion of successful prediction does not reach 90%, the above optimized neural network model is further optimized until the proportion of successful prediction reaches 90%.

[0072] It should be noted that the above settings of 5% error and 90% ratio are both optional settings, and the specific values ​​can be set according to actual conditions, and the present invention is not limited thereto.

[0073] S507: Determine an idleness prediction result of the target charging station according to the first label corresponding to each target charging pile included in the target charging station and a pre-trained prediction model.

[0074] See above. First, based on the first label corresponding to each target charging pile contained in the target charging station and the prediction model, the idleness prediction results of all target charging piles are obtained. Then, the idleness prediction results of all target charging piles are averaged to obtain the idleness prediction results of the target charging station. It should be noted that after the sample data is obtained above, the model training is performed after the sample data is standardized. Therefore, after the server obtains the first label corresponding to the target charging pile, it first standardizes the first label to obtain the standardized first label, and then inputs the standardized first label into the prediction model to obtain the idleness prediction result of the target charging pile.

[0075] The charging station idleness prediction method provided in this embodiment introduces the process of training a prediction model. Since the impact of multiple dimensions on idleness, such as the scale of the charging station, the evaluation score of the charging pile, the charging price, the charging mode, and the parking fee information, are taken into account during the training of the prediction model, the idleness prediction result obtained by the prediction model is more accurate.

[0076] Figure 6 This is a schematic diagram of the structure of the charging station idleness prediction device provided by the present invention. Figure 6As shown, the charging station idleness prediction device provided by the present invention includes:

[0077] The receiving module 601 is used to receive an idleness query request triggered by a user, wherein the idleness query request includes: an identifier of a target charging station requested to be queried by the user;

[0078] The acquisition module 602 is used to acquire, according to the identifier of the target charging station, a first tag corresponding to each target charging pile contained in the target charging station, wherein the first tag includes at least one of the following tags: the scale of the target charging station, the evaluation score of the target charging pile, the charging price of the target charging pile, the charging mode of the target charging pile, the parking fee information of the parking space corresponding to the target charging pile, the predicted time period, the week attribute corresponding to the predicted date, or the holiday attribute corresponding to the predicted date;

[0079] A determination module 603 is used to determine the idleness prediction result of the target charging station according to the first label corresponding to each target charging pile included in the target charging station and the pre-trained prediction model;

[0080] The output module 604 is used to output the idleness prediction result.

[0081] Optionally, the above device further includes: a training module 605; the training module 605 is used to:

[0082] Obtain training sample set and verification sample set;

[0083] After initializing the weights corresponding to the labels in the neural network model, the neural network model is trained using the training sample set to obtain an optimized neural network model;

[0084] The optimized neural network model is verified using the verification sample set. If the proportion of successfully verified sample data in the verification sample set exceeds a preset value, the optimized neural network model is determined as the prediction model.

[0085] Optionally, the training module 605 is specifically used for:

[0086] Get the historical idleness of any charging pile every hour within the preset time range;

[0087] Obtain a second tag corresponding to the charging pile for each hour within the preset time range, the second tag including at least one of the following tags: the scale of the charging station where the charging pile is located, the evaluation score of the charging pile, the charging price of the charging pile, the charging mode of the charging pile, the parking fee information of the parking space corresponding to the charging pile, the time period, the week attribute or the holiday attribute;

[0088] The historical idleness of the charging pile in any hour and the corresponding second tag are determined as a sample data.

[0089] Optionally, the training module 605 is specifically used for:

[0090] Obtain status information of any charging pile every hour within a preset time range, the status information including: idle, off-grid, faulty and / or plug-in charging status, the plug-in charging status including plug-in not charging or plug-in charging;

[0091] The historical idleness is determined according to the status information of each hour.

[0092] Optionally, the training module 605 is specifically used for:

[0093] Standardizing the label of each sample data in the N sample data;

[0094] The hash splitting algorithm is used to divide the standardized N sample data into a training group and a validation group.

[0095] Optionally, the determining module 603 is specifically configured to:

[0096] Obtaining idleness prediction results of all target charging piles according to the first tag corresponding to each target charging pile included in the target charging station and the prediction model;

[0097] The idleness prediction results of all the target charging piles are averaged to obtain the idleness prediction result of the target charging station.

[0098] Optionally, the neural network model includes: an input layer, two hidden layers and an output layer, the activation function of the hidden layer is a linear rectification function ReLU, and the activation function of the output layer is a linear function Liner.

[0099] Optionally, the determining module 603 is further configured to:

[0100] Storing the idleness prediction result of the target charging station;

[0101] receiving a query request triggered by a user for querying the availability of the target charging station;

[0102] According to the query request, the idleness prediction result of the target charging station is retrieved.

[0103] The charging station idleness prediction device provided in this embodiment can be used to execute the steps executed by the server in any of the above method embodiments. Its implementation principle and technical effects are similar and will not be repeated here.

[0104] The present invention provides a charging pile platform, on which functions such as charging station query and idleness query can be realized. A computer program is stored on the charging pile platform, and when the computer program is executed by a processor, the charging station idleness prediction method provided in the above embodiment can be realized.

[0105] The present invention provides a terminal device, which includes the above charging pile platform. The terminal device can be a mobile phone or a car computer.

[0106] Figure 7 The hardware structure diagram of the server provided by the present invention is shown in FIG. Figure 7 As shown, the server of this embodiment may include:

[0107] The memory 701 is used to store program instructions.

[0108] The processor 702 is used to implement the steps performed by the server in any of the above method embodiments when the program instructions are executed. The specific implementation principles can be found in the above embodiments, and will not be repeated in this embodiment.

[0109] The present invention provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps executed by the server in any of the above method embodiments are implemented.

[0110] The present invention also provides a program product, which includes a computer program stored in a readable storage medium. At least one processor can read the computer program from the readable storage medium, and the at least one processor executes the computer program so that the server implements the steps performed by the server in any of the above method embodiments.

[0111] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as 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.

[0112] 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.

[0113] In addition, each functional unit in each embodiment of the present invention 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. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0114] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (English: processor) to perform some steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (English: Read-Only Memory, abbreviated: ROM), random access memory (English: Random Access Memory, abbreviated: RAM), disk or optical disk and other media that can store program codes.

[0115] It should be understood that the processor described in the present invention may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the present application may be directly implemented as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.

[0116] 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 aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the idleness of a charging station, characterized in that: include: Receiving an availability query request triggered by a user clicking on an icon of a charging station distributed on a map, the availability query request including: an identifier of a target charging station, the map being displayed in an application software installed on a terminal device for querying the availability of charging stations; According to the identifier of the target charging station, a first tag corresponding to each target charging pile included in the target charging station is obtained, where the first tag includes at least two of the following tags: the scale of the target charging station, the evaluation score of the target charging pile, the charging price of the target charging pile, the charging mode of the target charging pile, the parking fee information of the parking space corresponding to the target charging pile, the predicted time period, the week attribute corresponding to the predicted date, or the holiday attribute corresponding to the predicted date; According to the dimension of the first label corresponding to each target charging pile included in the target charging station, a pre-trained prediction model applicable to the first label is selected to determine the idleness prediction result of the target charging station, wherein there are multiple pre-trained prediction models, and each prediction model uses a different label dimension when training; Before the method of selecting a pre-trained prediction model applicable to the first label according to the dimension of the first label corresponding to each target charging pile included in the target charging station and determining the idleness prediction result of the target charging station includes: Obtain training sample set and verification sample set; The obtaining of the training sample set and the verification sample set includes: Get N sample data, where N is a positive integer; The obtaining of N sample data includes: Get the historical idleness of any charging pile every hour within the preset time range; Obtain a second tag corresponding to the charging pile for each hour within the preset time range, the second tag including at least one of the following tags: the scale of the charging station where the charging pile is located, the evaluation score of the charging pile, the charging price of the charging pile, the charging mode of the charging pile, the parking fee information of the parking space corresponding to the charging pile, the time period, the week attribute or the holiday attribute; The historical idleness of the charging pile in any hour and the corresponding second tag are determined as a sample data.

2. The method according to claim 1, characterized in that After obtaining the training sample set and the verification sample set, the method further includes: After initializing the weights corresponding to the labels in the neural network model, the neural network model is trained using the training sample set to obtain an optimized neural network model; The optimized neural network model is verified using the verification sample set. If the proportion of successfully verified sample data in the verification sample set exceeds a preset value, the optimized neural network model is determined as the prediction model.

3. The method according to claim 2, characterized in that The obtaining of the training sample set and the verification sample set also includes: After obtaining N sample data, the N sample data are divided into a training group and a verification group using a hash splitting algorithm; Performing uniformity check on the training group and the verification group to obtain a uniformity check result; The training group and the validation group are adjusted respectively according to the uniformity check result to obtain the training sample set and the validation sample set.

4. The method according to claim 1, characterized in that: The step of obtaining the historical idleness of any charging pile every hour within a preset time range includes: Obtaining status information of any charging pile every hour within a preset time range, the status information including: idle, off-grid, faulty and / or plug-in charging status, the plug-in charging status including plug-in not charging or plug-in charging; The historical idleness is determined according to the status information of each hour.

5. The method according to claim 3, characterized in that: The method of using a hash splitting algorithm to divide the N sample data into a training group and a verification group includes: Standardizing the label of each sample data in the N sample data; The hash splitting algorithm is used to divide the standardized N sample data into a training group and a validation group.

6. The method according to any one of claims 1 to 4, characterized in that: The step of selecting a pre-trained prediction model applicable to the first label according to the dimension of the first label corresponding to each target charging pile included in the target charging station, and determining the idleness prediction result of the target charging station includes: Obtaining idleness prediction results of all target charging piles according to the first tag corresponding to each target charging pile included in the target charging station and the prediction model; The idleness prediction results of all the target charging piles are averaged to obtain the idleness prediction result of the target charging station.

7. The method according to any one of claims 2-3, characterized in that: The neural network model includes: an input layer, two hidden layers and an output layer. The activation function of the hidden layer is a linear rectifier function ReLU, and the activation function of the output layer is a linear function Liner.

8. The method according to any one of claims 1 to 4, characterized in that: The method further comprises: Storing the idleness prediction result of the target charging station; receiving a query request triggered by a user for querying the availability of the target charging station; According to the query request, the idleness prediction result of the target charging station is retrieved.

9. A charging pile platform, characterized in that: A computer program is stored on the charging pile platform, and when the computer program is executed by a processor, the method described in any one of claims 1 to 8 is implemented.

10. A terminal device, characterized in that: Including the charging pile platform described in claim 9.

Images