Electric vehicle load prediction method and device, electronic equipment and medium

Abstract

The invention relates to an electric vehicle load prediction method and device, electronic equipment and a medium, and the method comprises the steps: obtaining the travel data of an electric vehicle user, calculating the place transition probability of the electric vehicle based on the travel data, and determining the travel sequence of the user based on the place transition probability of the electric vehicle; the travel sequence of the user is combined with preset map data, and the predicted driving power consumption of the electric vehicle is output; acquiring first weather data of the place where the electric vehicle is located in the past 600 days, inputting the first weather data into the temperature prediction model which is completely trained, and outputting a predicted temperature value; the predicted temperature value is input into a preset temperature-air conditioner power consumption model, and air conditioner predicted power consumption of the electric vehicle is output; and combining the driving predicted power consumption with the air conditioner predicted power consumption, predicting the load of the electric vehicle, and outputting a prediction result. The load prediction of the electric vehicle is divided into the driving prediction and the air conditioner prediction, so that the load prediction precision of the electric vehicle is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric load forecasting, and in particular to an electric vehicle load forecasting method, device, electronic equipment and medium. Background Art

[0002] As the scale of the domestic electric vehicle market continues to expand, the difficulty of parking for electric vehicle users has become increasingly prominent, and the demand for charging is also growing; however, the demand for charging electric vehicles is highly random, and disordered charging behavior has a huge impact on the power grid. To this end, by leveraging the characteristics of load-side regulation resources such as electric vehicles and distributed energy storage, such as "flexible electricity consumption time, guideable electricity consumption behavior, predictable electricity consumption rules, and intelligent electricity consumption methods", they are incorporated into the power grid optimization control to fully tap the potential of electric vehicles and energy storage stations in the power market.

[0003] However, the charging behavior of electric vehicles is closely related to the user's work and rest habits. In order to avoid exacerbating the peak-to-valley difference in the power grid load, it is necessary to predict this part of the load, thereby assisting the dispatching behavior of the power grid. At present, there are two main types of electric vehicle load prediction: traditional prediction methods based on statistical methods and modern prediction methods based on machine learning. When predicting electric vehicle load, traditional prediction methods cannot predict the impact of factors such as weather on electric vehicle load; modern prediction methods cannot simulate and predict user travel behavior based on actual conditions, resulting in insufficient prediction accuracy and precision.

[0004] Therefore, how to provide an electric vehicle load forecasting method that can combine actual conditions and weather factors to improve the accuracy of electric vehicle load forecasting is an urgent problem to be solved. Summary of the invention

[0005] In view of this, it is necessary to provide an electric vehicle load prediction method, device, electronic device and medium to improve the accuracy of electric vehicle load prediction.

[0006] In order to achieve the above object, in a first aspect, the present invention provides an electric vehicle load forecasting method, comprising: Acquire travel data of an electric vehicle user, calculate the electric vehicle location transfer probability based on the travel data, and determine the user's travel sequence based on the electric vehicle location transfer probability; Combining the user's travel sequence with preset map data to output the predicted power consumption of the electric vehicle; Obtain the first weather data of the location of the electric vehicle in the past 600 days, wherein the first weather data includes temperature, humidity, air pressure, wind speed and direction, precipitation, cloud cover, visibility, UV index, air quality index, weather conditions, dew point temperature and relative humidity, input the first weather data into a well-trained temperature prediction model, and output a predicted temperature value; Inputting the predicted temperature value into a preset temperature-air conditioning power consumption model, and outputting the predicted air conditioning power consumption of the electric vehicle; The driving predicted power consumption and the air conditioning predicted power consumption are combined to predict the load of the electric vehicle and output the prediction result.

[0007] Furthermore, the calculating of the electric vehicle location transfer probability based on the travel data includes: Performing function fitting on the travel data to obtain a distribution curve of the number of electric vehicles transferred between different locations at different times; The electric vehicle location transfer probability is calculated based on the electric vehicle quantity distribution curve.

[0008] Furthermore, the user's travel sequence includes multiple trips, and the method of determining the user's travel sequence based on the electric vehicle location transfer probability includes: The destination of each trip is determined based on the electric vehicle location transfer probability.

[0009] Furthermore, the user's travel sequence also includes the starting and ending points of each trip, travel time, travel destination, driving speed and parking time information of each trip. The user's travel sequence is combined with the preset map data to output the predicted power consumption of the electric vehicle, including: Combining the starting and ending points of each segment of the travel sequence with the preset map data to obtain the travel distance of the electric vehicle in each segment of the travel sequence; Based on the driving distance of the electric vehicle in each trip, the predicted power consumption of the electric vehicle is determined:

[0010] in, Predicting the power consumption of electric vehicles, For electric vehicles The distance travelled during the trip, is the total number of segments of the electric vehicle’s journey, It is the average power consumption of electric vehicles per 100 kilometers.

[0011] Furthermore, the training process of the temperature prediction model includes: Acquire the second weather data of the location of the electric vehicle in the past 600 days, wherein the second weather data includes temperature, humidity, air pressure, wind speed and direction, precipitation, cloud cover, visibility, ultraviolet index, air quality index, weather conditions, dew point temperature and relative humidity, and pre-process the second weather data to obtain pre-processed weather data; Normalizing and encoding the preprocessed weather data to obtain encoded weather data; The encoded weather data is input into the initial temperature prediction model for parameter estimation, and after reaching the preset training termination condition, a fully trained temperature prediction model is obtained, wherein the initial temperature prediction model is composed of a convolutional neural network, a bidirectional long short-term memory neural network, a residual network, and an attention mechanism layer.

[0012] Further, the preprocessing the second weather data to obtain preprocessed weather data includes: Using a box plot to identify the second weather data, determining abnormal values ​​in the second weather data, and correcting the abnormal values ​​using a horizontal processing method combined with 8 data points to obtain preprocessed weather data, wherein the preprocessed weather data is numerical data; The step of normalizing and encoding the preprocessed weather data to obtain encoded weather data includes: The numerical data is linearly transformed using a preset data standardization method, and the transformed data is within a preset interval to obtain the encoded weather data.

[0013] Furthermore, the predicted temperature value is input into a preset temperature-air conditioning power consumption model to output the predicted air conditioning power consumption of the electric vehicle, including: Inputting the predicted temperature value into a preset temperature-air conditioning power consumption model, and outputting the air conditioning power consumption of the electric vehicle at the predicted temperature; Based on the air conditioning power consumption of the electric vehicle at the predicted temperature, the predicted air conditioning power consumption of the electric vehicle is calculated:

[0014] in, Predicting power consumption for electric vehicle air conditioning, For the The average travel speed during the trip. The power consumption of the car air conditioner at a temperature of 25°C. To predict the temperature, is the probability of using air conditioning in electric vehicles under the predicted temperature.

[0015] In a second aspect, the present invention further provides an electric vehicle load prediction device, comprising: A travel data acquisition module, used to acquire travel data of electric vehicle users, calculate the electric vehicle location transfer probability based on the travel data, and determine the user's travel sequence based on the electric vehicle location transfer probability; A driving power consumption prediction module, used to combine the user's travel sequence with preset map data and output the driving predicted power consumption of the electric vehicle; A weather data acquisition module is used to acquire the first weather data of the location of the electric vehicle in the past 600 days, wherein the first weather data includes temperature, humidity, air pressure, wind speed and direction, precipitation, cloud cover, visibility, UV index, air quality index, weather conditions, dew point temperature and relative humidity, and input the first weather data into a well-trained temperature prediction model to output a predicted temperature value; An air conditioning power consumption prediction module, which inputs the predicted temperature value into a preset temperature-air conditioning power consumption model, and outputs the predicted air conditioning power consumption of the electric vehicle; The electric vehicle load prediction module is used to combine the driving predicted power consumption and the air conditioning predicted power consumption, predict the load of the electric vehicle, and output the prediction result.

[0016] In a third aspect, the present invention further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps in the above-mentioned electric vehicle load prediction method when executing the computer program.

[0017] In a fourth aspect, the present invention further provides a computer storage medium, wherein the computer readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned electric vehicle load prediction method are implemented.

[0018] The beneficial effects of adopting the above embodiment are: The present invention divides the load of electric vehicles into two parts: driving power consumption and vehicle air conditioning power consumption, and uses specific prediction methods to predict them respectively, so that the load prediction of electric vehicles is more in line with the actual situation. First, the user's travel sequence is determined according to the user's travel data, wherein the travel sequence can simulate the user's travel behavior, and the mileage of the electric vehicle is accurately calculated in combination with the map data of the electric vehicle's location, so as to obtain the driving predicted power consumption; then, the local short-term weather conditions are predicted by training a complete temperature prediction model to obtain a predicted temperature value, and then the existing temperature-air conditioning power consumption model is combined to accurately calculate the predicted air conditioning power consumption of the electric vehicle, and finally the two parts of the load are superimposed to obtain a more accurate electric vehicle load prediction value. The prediction accuracy of electric vehicle load is improved, and it is beneficial to the economic operation and energy management of electric vehicle charging stations, and provides a reference basis for urban infrastructure planning and construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A schematic diagram of a flow chart of an embodiment of an electric vehicle load prediction method provided by the present invention; Figure 2 A topological diagram of a user's travel destination provided by an embodiment of the present invention; Figure 3 A schematic diagram of a temperature prediction model structure provided by an embodiment of the present invention; Figure 4 A schematic diagram of automobile air-conditioning power and air-conditioning usage rate at different temperatures provided by an embodiment of the present invention; Figure 5 A schematic diagram of driving routes of different users provided by an embodiment of the present invention; Figure 6 A distribution curve of the number of cars arriving at different locations at different times provided by an embodiment of the present invention; Figure 7 A user location transfer probability distribution diagram provided by an embodiment of the present invention; Figure 8 A schematic diagram of a prediction result after denormalization provided by an embodiment of the present invention; Fig. 9 A schematic diagram of load forecasting using different methods provided by an embodiment of the present invention; Fig.10 A schematic structural diagram of an embodiment of an electric vehicle load prediction device provided by the present invention; Fig.11 A schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0021] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. Reference to "embodiments" in this article means that the specific features, structures or characteristics described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0022] The specific embodiments are described in detail below: See also Figure 1 , Figure 1 A flow chart of an embodiment of an electric vehicle load prediction method provided by the present invention is provided. A specific embodiment of the present invention discloses an electric vehicle load prediction method, comprising: Step S101: Acquire travel data of an electric vehicle user, calculate the electric vehicle location transfer probability based on the travel data, and determine the user's travel sequence based on the electric vehicle location transfer probability; Step S102: combining the user's travel sequence with preset map data to output the predicted power consumption of the electric vehicle; Step S103: obtaining first weather data of the location of the electric vehicle in the past 600 days, wherein the first weather data includes temperature, humidity, air pressure, wind speed and direction, precipitation, cloud cover, visibility, ultraviolet index, air quality index, weather conditions, dew point temperature and relative humidity, inputting the first weather data into a well-trained temperature prediction model, and outputting a predicted temperature value; Step S104: inputting the predicted temperature value into a preset temperature-air conditioning predicted power consumption model, and outputting the predicted air conditioning power consumption of the electric vehicle; Step S105: Combine the predicted driving power consumption and the predicted air conditioning power consumption to predict the load of the electric vehicle and output the prediction result.

[0023] The present invention divides the load of electric vehicles into two parts: driving power consumption and vehicle air conditioning power consumption, and uses specific prediction methods to predict them respectively, so that the load prediction of electric vehicles is more in line with the actual situation. First, the user's travel sequence is determined according to the user's travel data, wherein the travel sequence can simulate the user's travel behavior, and the mileage of the electric vehicle is accurately calculated in combination with the map data of the electric vehicle's location, so as to obtain the driving predicted power consumption; then, the local short-term weather conditions are predicted by training a complete temperature prediction model to obtain a predicted temperature value, and then the existing temperature-air conditioning power consumption model is combined to accurately calculate the predicted air conditioning power consumption of the electric vehicle, and finally the two parts of the load are superimposed to obtain a more accurate electric vehicle load prediction value. The prediction accuracy of electric vehicle load is improved, and it is beneficial to the economic operation and energy management of electric vehicle charging stations, and provides a reference basis for urban infrastructure planning and construction.

[0024] In one embodiment of the present invention, calculating the electric vehicle location transfer probability based on travel data includes: Function fitting is performed on travel data to obtain the distribution curve of the number of electric vehicles transferred between different locations at different times; The probability of electric vehicle location transfer is calculated based on the electric vehicle number distribution curve.

[0025] It is understandable that the travel data of electric vehicle users includes but is not limited to the travel location, stay time, driving speed, departure time, end time, distance between two places, driving time of the trip and parking time at the destination, etc., and satisfies the following relationship: , , .in, Indicates the location number. Indicates the parking time. Indicates the driving time between different destinations. and Indicates arrival and departure respectively The time of the location, Indicates the distance between the electric vehicle’s current location and the next destination. Represents the average driving speed of electric vehicles in the area.

[0026] Function fitting is performed on travel data to obtain the user's spatiotemporal transfer probability curve, specifically the distribution curve of the number of electric vehicles transferred between different locations at different times; Among them, the fitting process of the function fitting method is as follows: Assume that user travel data can be described by the following function:

[0027] In the formula, the parameter to be estimated is , are the two peaks of the curve to be fitted, , are the peak positions, , are the half widths of the two peaks of the curve, represents a data point monitored in time, express The number of vehicles at the time. Considering all user travel data, the parameters to be estimated are calculated according to the matrix calculation principle to obtain the time-space transfer probability curve that meets the user, specifically the function characteristic parameters of the distribution curve of the number of electric vehicles transferred between different locations at different times. Based on the fitting, the first travel time of the user at a certain location can be obtained, and the difference in the first travel time between any two destinations can be used to calculate the parking time at a certain location.

[0028] The electric vehicle location transfer probability represents the probability of going from the current location to the next location, which is usually related to the current time and the current location. The electric vehicle location transfer probability is calculated by fitting the travel data of users from the current location to different functional areas using the following formula: ,in Indicates that the sample The number of cars at point i going to point j at a given moment, express Time from Possible destinations include indivual.

[0029] In one embodiment of the present invention, the user's travel sequence includes multiple trips, and determining the user's travel sequence based on the electric vehicle location transfer probability includes: The destination of each trip is determined based on the electric vehicle location transfer probability.

[0030] It is understandable that due to the different travel needs of different users, different travel needs are divided into multiple trips, and the arrival point of the previous trip is the same as the departure point of the next trip. The travel sequence of a certain user can be formed based on the simulated starting point, travel time, travel destination, driving speed, and parking time information of each trip, and the user's next travel destination can be determined by the electric vehicle location transfer probability.

[0031] In one embodiment of the present invention, the user's travel sequence includes the starting and ending points of each trip, the travel time, the travel destination, the driving speed and the parking time information of each trip. The user's travel sequence is combined with the preset map data to output the predicted power consumption of the electric vehicle, including: The starting and ending points of each segment of the travel sequence are combined with the preset map data to obtain the driving distance of the electric vehicle in each segment; Based on the driving distance of the electric vehicle in each trip, the predicted power consumption of the electric vehicle is determined:

[0032] in, Predicting the power consumption of electric vehicles, For electric vehicles The distance travelled during the trip, is the total number of segments of the electric vehicle’s journey, It is the average power consumption of electric vehicles per 100 kilometers.

[0033] It is understandable that the user's travel sequence is divided into multiple trips, and the arrival point of the previous trip is the same as the departure point of the next trip. The user's travel sequence includes the starting and ending points of each trip, travel time, travel destination, driving speed and parking time information of each trip. Therefore, when constructing a user's complex travel sequence, the owner's next travel destination is determined by the location transfer probability of the above method, combined with the map distance information in the preset map data and the average driving speed of local car owners, to simulate the complete travel information of the owner, so as to determine the driving power consumption in the process: ,in, Predicting the power consumption of electric vehicles, For electric vehicles The distance travelled during the trip, is the total number of segments of the electric vehicle’s journey, The average power consumption of electric vehicles per 100 kilometers.

[0034] Among them, the preset map data can be divided according to the main functional areas of the city, and the travel destinations can be divided into five categories according to people's daily travel needs: residential place H, work place W, catering and shopping place SE, entertainment place SR and other places R, abstracting and simplifying the city map into a user travel destination topology map, such as Figure 2 As shown, Figure 2 A user travel destination topology map provided by an embodiment of the present invention.

[0035] In one embodiment of the present invention, the training process of the temperature prediction model includes: Obtain the second-day weather data of the location of the electric vehicle in the past 600 days, wherein the second-day weather data includes temperature, humidity, air pressure, wind speed and direction, precipitation, cloud cover, visibility, UV index, air quality index, weather conditions, dew point temperature and relative humidity, and pre-process the second-day weather data to obtain pre-processed weather data; Normalizing and encoding the preprocessed weather data to obtain encoded weather data; The encoded weather data is input into the initial temperature prediction model for parameter estimation. After reaching the preset training termination condition, a fully trained temperature prediction model is obtained, wherein the initial temperature prediction model is composed of a convolutional neural network, a bidirectional long short-term memory neural network, a residual network, and an attention mechanism layer.

[0036] First of all, it should be noted that for electric vehicle users, the energy consumption of the car air conditioner is the second largest power consumption module in the entire vehicle use process. Therefore, weather factors, especially temperature, are also particularly important for electric vehicle load prediction. Therefore, the prediction of air conditioning power consumption in the present invention can start from temperature. Specifically, the prediction of air conditioning power consumption can adopt the method of neural network model to predict temperature, and finally use the temperature value to reversely calculate the car air conditioning power consumption.

[0037] Among them, the training process of the temperature prediction model is as follows: First, the second weather data such as temperature, humidity, air pressure, wind speed and direction, precipitation, cloud cover, visibility, ultraviolet index, air quality index, weather conditions, dew point temperature, relative humidity, etc. of the location of the electric vehicle in the past 600 days are obtained, and the second weather data are preprocessed to obtain preprocessed weather data; wherein, the preprocessing of the second weather data includes: using a box plot to identify the second weather data, determining the abnormal values ​​in the second weather data, and using a horizontal processing method combined with 8 data points to correct the abnormal values, and obtaining the preprocessed weather data, wherein the preprocessed weather data is numerical data; It is understandable that when preprocessing the second-day weather data, since weather data may be missing, wrongly recorded, etc., if these data are used directly without processing, it may lead to a decrease in the accuracy of model prediction. Therefore, data preprocessing is required before prediction. The present invention uses a box plot to identify and screen abnormal data. The box plot can clearly determine the abnormal values ​​in each group of data. The time interval of the data used in the present invention is 15 minutes. The weather conditions usually do not change dramatically at a shorter time interval. Therefore, the horizontal processing method can be used to process and replace the identified abnormal values, that is, the load average value of the 4 moments before and after the abnormal point is taken as the correction value of the point. The processing method is as follows:

[0038] in, is the abnormal data location, is the data correction value at this position, , , , is the 1st to 4th data value before this position, , , , It is the 1st to 4th data value after this position. The preprocessed weather data is numerical data.

[0039] Then the preprocessed data is normalized and encoded to obtain the encoded weather data.

[0040] It is understandable that due to the different measurement methods in weather data, there are usually differences in magnitude. In order to avoid convergence difficulties and reduced prediction accuracy caused by differences in magnitude, these data need to be normalized. In one embodiment of the present invention, the pre-processed weather data is normalized and encoded to obtain encoded weather data, including: The numerical data is linearly transformed using a preset data standardization method, and the transformed data is within a preset range to obtain encoded weather data.

[0041] Specifically, the data standardization method is used to perform linear transformation on the numerical data to the interval [0, 1]. The formula is: ,in, is the standardized data, Represents the original data, and Respectively represent the minimum and maximum values ​​in a certain weather data.

[0042] The encoded weather data is then input into the initial temperature prediction model for parameter estimation. After the preset training termination condition is reached, a fully trained temperature prediction model is obtained. Figure 3 As shown, Figure 3 A schematic diagram of the structure of an initial temperature prediction model provided for an embodiment of the present invention is provided. The initial temperature prediction model is composed of a convolutional neural network, a bidirectional long short-term memory neural network, a residual network, and an attention mechanism layer, wherein the convolutional neural network is mainly responsible for feature extraction. First, a sequence folding layer is set to fold the data; then three convolutional layers and two pooling layers are designed, and a three-dimensional array is obtained after three consecutive convolutions and pooling, which is used as the input of the bidirectional long short-term memory neural network after sequence expansion and flattening; the bidirectional long short-term memory neural network is mainly responsible for load prediction, and the prediction accuracy is related to the number of its layers and the number of hidden units therein; the residual network solves the gradient vanishing problem in deep network training by introducing jump connections, and the residual module allows the network to learn the residual between input and output, which helps to improve the performance of the deep network; the attention mechanism layer enables the model to focus on the most important part of the time series, helping the model to identify which time points are most critical for predicting future temperatures, thereby improving the accuracy of the prediction.

[0043] In order to evaluate the quality of the prediction method, the residual sum of squares, R square, mean square error, root mean square error, square absolute error, etc. can usually be used for evaluation. In one embodiment of the present invention, the average relative error is selected: And the root mean square error: To evaluate, where N is the number of prediction points, is the actual value, For the predicted value, when the prediction result MAPE<5% and RMSE<3, it indicates that the error is acceptable.

[0044] It should also be noted that since the weather forecast value does not have dimensions after processing and forecasting, it needs to be processed before outputting the forecast result to obtain dimensioned data that conforms to reality. The processing method is as follows: ,in, is the processed prediction data, is the predicted value before processing.

[0045] In one embodiment of the present invention, the predicted temperature value is input into a preset temperature-air conditioning power consumption model, and the predicted air conditioning power consumption of the electric vehicle is output, including: Input the predicted temperature value into a preset temperature-air conditioning power consumption model, and output the air conditioning power consumption of the electric vehicle at the predicted temperature; Based on the predicted air conditioning power consumption of electric vehicles at the predicted temperature, the predicted air conditioning power consumption of electric vehicles is calculated:

[0046] in, Predicting power consumption for electric vehicle air conditioning, For the The average travel speed during the trip. The power consumption of the car air conditioner at a temperature of 25°C. To predict the temperature, is the probability of using air conditioning in electric vehicles under the predicted temperature.

[0047] It is understandable that the preset temperature-air conditioning power consumption model can refer to Figure 4 , Figure 4 This is a schematic diagram of the air conditioning power and air conditioning usage rate of an automobile at different temperatures provided by an embodiment of the present invention. Therefore, by inputting the predicted temperature value, the current user's air conditioning usage and air conditioning power consumption can be obtained. After obtaining the air conditioning power consumption of the electric vehicle at the predicted temperature, it can be further calculated according to the formula The predicted power consumption of the air conditioner of the electric vehicle is obtained. Finally, the predicted power consumption of driving and the predicted power consumption of air conditioner are combined to predict the load of the electric vehicle and output the prediction result: .

[0048] It should also be noted that since the charging behavior of users is highly subjective, it is assumed that as long as there are charging facilities at each parking area, they will immediately participate in charging until they leave the area. Considering the increasingly complete charging facilities in the current urban area and the emergence of multi-functional parking stations, charging facilities are set at four locations: H, W, SR, and SE. Charging facilities are not set at other locations. The charging mode is set to L for slow charging and F for fast charging. Therefore, if the charging power at a certain location is sufficient, .in, Indicates the car battery capacity, Indicates the remaining battery power of the car when arriving at the location. Indicates the charging efficiency, Indicates the charging power, Indicates the time the car stays at this point. Since the vehicles of ordinary car users are in a parking state in most cases, they can be regarded as being in a charging state, so the situation where the electric car is not charged enough does not need to be considered in the present invention.

[0049] In order to understand the present invention more clearly, the embodiment of the present invention also provides an electric vehicle load prediction scheme, which selects a certain brand of car with vehicle parameters of 45KWh capacity, 7KW slow charging power, 30KW fast charging power, and 0.9 charging efficiency, and takes the typical summer weather in a certain area as an example to predict the travel behavior of electric vehicle users in the area and the electric vehicle load. The prediction mainly includes two parts: predicting driving power consumption based on the travel sequence method and predicting air conditioning power consumption based on the neural network model. The whole process includes the following steps: Regarding the prediction of driving power consumption: Collect map data of the area. First, divide the travel destinations into the following five categories according to people's daily travel needs: residential place H, work place W, dining and shopping place SE, entertainment place SR and other places R. Retain information elements such as the distance between functional areas to simplify the map, forming a simplified map, such as Figure 5 As shown, Figure 5 A schematic diagram of driving routes for different users provided in one embodiment of the present invention. The distance between any two functional areas obtained according to the actual distance measurement value is shown in Table 1: Table 1 Travel location distance values

[0050] The user's travel chain information includes the departure time, end time, distance between two places, driving time and parking time at the destination of each trip, and satisfies the following relationship: , , Among them, i means that you are at location i at this time. Indicates the parking time. Indicates driving time. and They represent the time points of arrival and departure at location i respectively. The driving speed is 65 km / h on a sunny day and 30 km / h on a rainy day. Since the parking time is highly subjective, a random method can be used for simulation during the prediction.

[0051] By investigating the travel data of local car users, the user's time-space distribution curve is fitted using a function. The number of vehicles arriving at a certain place at each time in a day is counted at 15-minute intervals to represent the user's travel tendency, such as Figure 6 As shown, Figure 6 A distribution curve of the number of cars arriving at different places at different times is provided in one embodiment of the present invention. According to the travel data of 300 car users in the area, the fitting time-space distribution curve, that is, the distribution curve of the number of cars arriving at different places at different times is: H: 、W: 、SE: 、SR: 、R: .

[0052] It can be understood that the location transfer probability represents the probability of going from the current location to the next location, which is usually related to the current time and the current location of the car. Taking the departure from location W as an example, Figure 7 As shown, Figure 7 A user location transfer probability distribution diagram is provided in an embodiment of the present invention. The calculation formula of the user location transfer probability is: ,in, represents the number of cars in the sample that are located at location i and go to location j at time t. According to the fitting results, it can be seen that between 12:00 and 1:00 noon, there are about 2,000 cars going from work locations to dining and shopping areas, about 250 users going to entertainment locations, and about 200 car owners choosing to go to other locations. Therefore, the transfer probability starting from location W to other locations can be calculated as: , , The calculation methods of other transfer probabilities are similar and will not be repeated here.

[0053] Due to the different travel needs of different users, different travel needs are divided into multiple trips, and the arrival point of the previous trip is the same as the departure point of the next trip. The travel sequence of a certain user can be formed based on the simulated starting point, travel time, travel destination, driving speed, and parking time of each trip. The user's next travel destination is determined by the user's location transfer probability, combined with the simplified map distance information, to obtain the user's driving distance today, and finally determine the driving power consumption in the process by the following formula: .in, is the distance traveled by the vehicle in a certain journey, The average power consumption of the car per 100 kilometers is 15 kWh / 100 kilometers.

[0054] For example, Figure 5 The simulated driving routes of two users in a day. User A starts work from home at 8:30 in the morning, arrives at the company at around 9:00, goes to a restaurant for lunch at 12:00, goes to other places for other affairs at 13:00, and finally returns home at 16:00. His driving route for the day is H1-W-SE / SR3-R1-H1. User B also starts work at 8:30 in the morning, has lunch and rest at the company, and returns home after get off work at 17:30. His driving route for the day is H2-W-H2. Combined with the previous analysis, we can predict the daily driving consumption of the two users respectively: A: 、B: .

[0055] Regarding the prediction of air conditioning power consumption: The embodiment of the present invention selects the actual weather data of the region in the first 15 days of August as a data set, and divides the training set and the test set into a ratio of 8:2 for training and prediction. A day is divided into 24 time periods, and the retained weather data includes the outdoor temperature, air quality index, and air humidity at each time point. Since weather data may be missing, misrecorded, or omitted, data preprocessing is required before prediction. The present invention uses a box plot to identify and filter the abnormal data. The abnormal values ​​in each group of data can be clearly judged through the box plot. Since the time interval of the data used in the statistics of air conditioning power consumption in this embodiment is 15 minutes, the load will not change drastically compared to the 1-hour time interval of the weather forecast, so the load curve presents a smooth and continuous state.

[0056] Since the outdoor temperature, air quality index, and air humidity data in the weather data have order of magnitude differences, these data need to be normalized to avoid convergence difficulties and reduced prediction accuracy caused by order of magnitude differences. The normalization method is used to linearly transform the numerical data to the interval [0, 1]. The formula is as follows: ,in, represents the standardized data, Represents the original data, and Respectively represent the minimum and maximum values ​​in a certain weather data.

[0057] like Figure 3 As shown in the figure, the initial temperature prediction model consists of a convolutional neural network, a bidirectional long short-term memory neural network, a residual network, and an attention mechanism layer. The convolutional neural network is mainly responsible for feature extraction. First, a sequence folding layer is set to fold the data; then three convolutional layers and two pooling layers are designed. After three consecutive convolutions and pooling, a three-dimensional array is obtained, which is used as the input of the BiLSTM network after sequence expansion and flattening; BiLSTM is mainly responsible for load prediction. Finally, after optimization, the number of neurons is determined to be 56, the number of iterations is 300, the learning rate is 0.0049, and the fitness effect is good.

[0058] In order to evaluate the model more objectively, the average relative error is selected: And the root mean square error: For evaluation, N is the number of prediction points, is the actual value, The smaller the predicted value, MAPE, and RMSE, the better the prediction effect. In the embodiment, the prediction error MAPE=1.9%, RMSE=0.69 degrees Celsius, and the error between the prediction and the actual value is small, which can meet the prediction error requirement. It should also be noted that since the weather prediction value does not have a dimension after processing and prediction, it needs to be denormalized before the weather prediction result can be output. Figure 8 As shown, Figure 8 A schematic diagram of a prediction result after denormalization provided by an embodiment of the present invention. The denormalization formula is as follows: ,in, is the predicted data after denormalization, is the predicted value before denormalization.

[0059] Existing research on automobile air conditioning power and air conditioning usage rate at different temperatures is as follows Figure 4 As shown, based on the weather temperature value predicted in the previous step, the probability that the owner will choose to turn on the air conditioner at the current temperature and the power consumption of the air conditioner at this temperature can be estimated, that is: ,in, Predicting power consumption for electric vehicle air conditioning, For the The average travel speed during the trip. For Predict the power consumption of car air conditioners at all times and at different temperatures. To predict the temperature, is the probability of using air conditioning in electric vehicles under the predicted temperature.

[0060] Taking user A and user B as an example, based on the simulated driving routes of the two users for one day and the local temperature forecast, query Figure 4 The air conditioning power consumption of two users in one day can be calculated based on the car air conditioning power and air conditioning usage rate at different temperatures. Among them, the air conditioning power consumption of user A in one day is: ; The air conditioning power consumption of user B in one day is: ; The combined load forecast of the two electric vehicles is: A: 、B: .

[0061] In order to better demonstrate the effect achieved by the present invention, this embodiment predicts the driving power consumption and air conditioning power consumption of 300 vehicles in the area in one day, and compares the prediction using the traditional Monte Carlo method with the method adopted by the present invention. The results are shown in the attached figure. Fig. 9 As shown, Fig. 9 A schematic diagram of load forecasting using different methods provided by an embodiment of the present invention.

[0062] Depend on Fig. 9 It can be seen that due to the overall high temperature in the region in summer, the prediction results in the present invention are also significantly higher than the traditional Monte Carlo method. However, at 4-6 in the morning and 4-6 in the afternoon, due to the change in temperature during the alternation of day and night, the usage rate of the air conditioner is significantly reduced, and the overall load of the vehicle is closer to the traditional prediction method; in other time periods, due to the overall decline in demand for automobile use, the prediction effects of the traditional method and the method of the present invention are not much different. It can be seen that compared with the traditional Monte Carlo simulation method, the method of the present invention divides the electric vehicle load into two parts, namely, driving power consumption and vehicle air conditioning power consumption, and uses specific prediction methods for prediction respectively, so that the electric vehicle load prediction is more in line with the actual situation, and can provide certain data references for future load scheduling and electric vehicle charging station construction.

[0063] In order to better implement the electric vehicle load prediction method in the embodiment of the present invention, based on the electric vehicle load prediction method, correspondingly, please refer to Fig.10 , Fig.10 This is a schematic diagram of the structure of an embodiment of an electric vehicle load prediction device provided by the present invention. The embodiment of the present invention provides an electric vehicle load prediction device 1000, including: The travel data acquisition module 1001 is used to acquire the travel data of the electric vehicle user, calculate the electric vehicle location transfer probability based on the travel data, and determine the user's travel sequence based on the electric vehicle location transfer probability; The driving power consumption prediction module 1002 is used to combine the user's travel sequence with the preset map data and output the driving predicted power consumption of the electric vehicle; The weather data acquisition module 1003 is used to acquire the first weather data of the location of the electric vehicle in the past 600 days, wherein the first weather data includes temperature, humidity, air pressure, wind speed and direction, precipitation, cloud cover, visibility, UV index, air quality index, weather conditions, dew point temperature and relative humidity, and input the first weather data into a well-trained temperature prediction model to output a predicted temperature value; The air conditioning power consumption prediction module 1004 inputs the predicted temperature value into a preset temperature-air conditioning power consumption model, and outputs the predicted air conditioning power consumption of the electric vehicle; The electric vehicle load prediction module 1005 is used to combine the driving power consumption and the air conditioning power consumption to predict the load of the electric vehicle and output the prediction result.

[0064] It should be noted here that: the device 1000 provided in the above embodiment can implement the technical solutions described in the above method embodiments. The specific implementation principles of the above modules or units can refer to the corresponding contents in the above method embodiments, which will not be repeated here.

[0065] Based on the above-mentioned electric vehicle load prediction method, an embodiment of the present invention also provides an electronic device accordingly, including: a processor and a memory, and a computer program stored in the memory and executable on the processor; when the processor executes the computer program, the steps of the electric vehicle load prediction method in the above-mentioned embodiments are implemented.

[0066] Fig.11 1 shows a schematic diagram of the structure of an electronic device 1100 suitable for implementing an embodiment of the present invention. The electronic device in the embodiment of the present invention may include, but is not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Fig.11 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0067] The electronic device includes: a memory and a processor, wherein the processor here may be referred to as a processing device 1101 hereinafter, and the memory may include at least one of a read-only memory (ROM) 1102, a random access memory (RAM) 1103, and a storage device 1108 hereinafter, as specifically shown below: like Fig.11 As shown, the electronic device 1100 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 1101, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1102 or a program loaded from a storage device 1108 to a random access memory (RAM) 1103. In the RAM 1103, various programs and data required for the operation of the electronic device 1100 are also stored. The processing device 1101, the ROM 1102, and the RAM 1103 are connected to each other via a bus 1104. An input / output (I / O) interface 1105 is also connected to the bus 1104.

[0068] Typically, the following devices may be connected to the I / O interface 1105: an input device 1106 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 1107 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 1108 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1109. The communication device 1109 may allow the electronic device 1100 to communicate with other devices wirelessly or by wire to exchange data. Although Fig.11 The electronic device 1100 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0069] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 1109, or installed from the storage device 1108, or installed from the ROM 1102. When the computer program is executed by the processing device 1101, the above-mentioned functions defined in the method of the embodiment of the present invention are executed.

[0070] Based on the above-mentioned electric vehicle load prediction method, an embodiment of the present invention also provides a computer-readable storage medium, which stores one or more programs. The one or more programs can be executed by one or more processors to implement the steps in the electric vehicle load prediction method in the above-mentioned embodiments.

[0071] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0072] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for predicting electric vehicle load, characterized in that: include: Acquire travel data of an electric vehicle user, calculate the electric vehicle location transfer probability based on the travel data, and determine the user's travel sequence based on the electric vehicle location transfer probability; Combining the user's travel sequence with preset map data to output the predicted power consumption of the electric vehicle; Obtain the first weather data of the location of the electric vehicle in the past 600 days, wherein the first weather data includes temperature, humidity, air pressure, wind speed and direction, precipitation, cloud cover, visibility, UV index, air quality index, weather conditions, dew point temperature and relative humidity, input the first weather data into a well-trained temperature prediction model, and output a predicted temperature value; Inputting the predicted temperature value into a preset temperature-air conditioning power consumption model, and outputting the predicted air conditioning power consumption of the electric vehicle; The driving predicted power consumption and the air conditioning predicted power consumption are combined to predict the load of the electric vehicle, and the prediction result is output.

2. The electric vehicle load forecasting method according to claim 1, characterized in that: The calculating the location transfer probability of the electric vehicle based on the travel data includes: Performing function fitting on the travel data to obtain a distribution curve of the number of electric vehicles transferred between different locations at different times; The electric vehicle location transfer probability is calculated based on the electric vehicle quantity distribution curve.

3. The electric vehicle load forecasting method according to claim 1, characterized in that: The user's travel sequence includes multiple trips, and the method of determining the user's travel sequence based on the electric vehicle location transfer probability includes: The destination of each trip is determined based on the electric vehicle location transfer probability.

4. The electric vehicle load forecasting method according to claim 3, characterized in that: The user's travel sequence also includes the starting and ending points of each trip, travel time, travel destination, driving speed and parking time information of each trip. The user's travel sequence is combined with the preset map data to output the predicted power consumption of the electric vehicle, including: Combining the starting and ending points of each segment of the travel sequence with the preset map data to obtain the travel distance of the electric vehicle in each segment of the travel sequence; Based on the driving distance of the electric vehicle in each trip, the predicted power consumption of the electric vehicle is determined: in, Predicting the power consumption of electric vehicles, For electric vehicles The distance travelled during the trip, is the total number of segments of the electric vehicle’s journey, It is the average power consumption of electric vehicles per 100 kilometers.

5. The electric vehicle load forecasting method according to claim 1, characterized in that: The training process of the temperature prediction model includes: Acquire the second weather data of the location of the electric vehicle in the past 600 days, wherein the second weather data includes temperature, humidity, air pressure, wind speed and direction, precipitation, cloud cover, visibility, ultraviolet index, air quality index, weather conditions, dew point temperature and relative humidity, and pre-process the second weather data to obtain pre-processed weather data; Normalizing and encoding the preprocessed weather data to obtain encoded weather data; The encoded weather data is input into the initial temperature prediction model for parameter estimation, and after reaching the preset training termination condition, a fully trained temperature prediction model is obtained, wherein the initial temperature prediction model is composed of a convolutional neural network, a bidirectional long short-term memory neural network, a residual network, and an attention mechanism layer.

6. The electric vehicle load forecasting method according to claim 5, characterized in that: The preprocessing of the second weather data to obtain preprocessed weather data includes: Using a box plot to identify the second weather data, determining abnormal values ​​in the second weather data, and correcting the abnormal values ​​using a horizontal processing method combined with 8 data points to obtain preprocessed weather data, wherein the preprocessed weather data is numerical data; The step of normalizing and encoding the preprocessed weather data to obtain encoded weather data includes: The numerical data is linearly transformed using a preset data standardization method, and the transformed data is within a preset interval to obtain the encoded weather data.

7. The electric vehicle load forecasting method according to claim 4, characterized in that: The step of inputting the predicted temperature value into a preset temperature-air conditioning power consumption model and outputting the predicted air conditioning power consumption of the electric vehicle comprises: Inputting the predicted temperature value into a preset temperature-air conditioning power consumption model, and outputting the air conditioning power consumption of the electric vehicle at the predicted temperature; Based on the air conditioning power consumption of the electric vehicle at the predicted temperature, the predicted air conditioning power consumption of the electric vehicle is calculated: in, Predicting power consumption for electric vehicle air conditioning, For the The average travel speed during the trip. The power consumption of the car air conditioner at a temperature of 25°C. To predict the temperature, is the probability of using air conditioning in electric vehicles under the predicted temperature.

8. An electric vehicle load prediction device, characterized in that: include: A travel data acquisition module, used to acquire travel data of electric vehicle users, calculate the electric vehicle location transfer probability based on the travel data, and determine the user's travel sequence based on the electric vehicle location transfer probability; A driving power consumption prediction module, used to combine the user's travel sequence with preset map data and output the driving predicted power consumption of the electric vehicle; A weather data acquisition module is used to acquire the first weather data of the location of the electric vehicle in the past 600 days, wherein the first weather data includes temperature, humidity, air pressure, wind speed and direction, precipitation, cloud cover, visibility, UV index, air quality index, weather conditions, dew point temperature and relative humidity, and input the first weather data into a well-trained temperature prediction model to output a predicted temperature value; An air conditioning power consumption prediction module, which inputs the predicted temperature value into a preset temperature-air conditioning power consumption model, and outputs the predicted air conditioning power consumption of the electric vehicle; The electric vehicle load prediction module is used to combine the driving predicted power consumption and the air conditioning predicted power consumption, predict the load of the electric vehicle, and output the prediction result.

9. An electronic device, characterized in that: It includes a memory and a processor, wherein the memory is used to store programs; the processor is coupled to the memory and is used to execute the programs stored in the memory to implement the steps in the electric vehicle load prediction method described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: Used to store computer-readable programs or instructions, which, when executed by a processor, can implement the steps in the electric vehicle load forecasting method described in any one of claims 1 to 7.

Images