Charging station load uncertainty quantification method based on space-time traffic flow prediction

Abstract

The invention discloses a charging station load uncertainty quantification method based on space-time traffic flow prediction, and the method comprises the steps: collecting traffic flow data from sensor nodes at two sides of a road, including traffic flow data and traffic flow average speed data; constructing a traffic flow uncertainty prediction model based on the quantile space-time network, and predicting the traffic flow uncertainty by using the model and the traffic flow data; considering the charging electricity price, traffic flow and road congestion factors of the charging station, an improved charging station queuing model is constructed, and the uncertainty prediction result of the traffic flow is converted into the uncertainty result of the load of the charging station. The method fully excavates the space-time coupling characteristics of the traffic flow data, effectively improves the accuracy of traffic flow uncertainty prediction, comprehensively considers the influence of the traffic flow, the charging electricity price and other factors on the charging station load, and further improves the precision of charging station load uncertainty prediction.

Description

Technical Field

[0001] The invention relates to a charging station load uncertainty quantification method based on spatiotemporal traffic flow prediction, and belongs to the field of power systems. Background Art

[0002] The rapid development of electric vehicles (EVs) is changing the global transportation landscape, with the core driving force being the urgent pursuit of carbon emission reduction and clean alternative energy. This trend has led to a sharp increase in the number of electric vehicles on the road and their charging needs, which in turn poses a severe challenge to the stability and power supply capacity of the power grid. In addition, the charging load of electric vehicles is highly non-stationary, and traditional deterministic predictions are difficult to accurately describe their load characteristics. Uncertainty prediction can quantify the uncertainty of electric vehicles and plays a vital supporting role in the dispatching and control decisions of the power grid.

[0003] Existing research on electric vehicle load prediction has not fully considered the impact of traffic flow, road congestion coefficient, electricity price, etc. on the queuing process. On the other hand, existing research on traffic flow uncertainty prediction cannot fully consider the spatiotemporal correlation between traffic flows at various charging station nodes, which restricts the accuracy of charging load uncertainty prediction. Therefore, in order to effectively improve the accuracy of charging station load uncertainty, the present invention proposes a charging station load uncertainty quantification method based on spatiotemporal traffic flow prediction. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method for quantifying the uncertainty of charging station load based on spatiotemporal traffic flow prediction. The present invention can fully consider the spatiotemporal correlation characteristics of traffic flow data to predict traffic flow uncertainty. On this basis, it can consider the charging electricity price, traffic flow and road congestion factors of the charging station, and realize the accurate conversion of traffic flow uncertainty to charging station load uncertainty. To achieve the above purpose, the present invention is implemented by the following technical solution, which specifically includes the following steps: (1) Collect traffic flow data from sensor nodes on both sides of the road, including traffic volume data and average vehicle speed data; (2) Construct and train a traffic flow uncertainty prediction model based on quantile spatiotemporal networks, and use this model and traffic flow data to predict traffic flow uncertainty; (3) Construct an electric vehicle queuing model for charging stations that takes into account charging electricity prices, traffic flow, and road congestion, and convert the traffic flow uncertainty prediction results into charging station load uncertainty prediction results.

[0005] Furthermore, the step (1) comprises: Traffic flow data is collected from sensor nodes on both sides of the road, including traffic flow data and average vehicle speed data, which can be expressed as ,in is the total traffic flow data, express The traffic flow data at that time, N is the number of nodes, T is the time length, H is the number of traffic flow data features of each node, and R represents a real number set.

[0006] Furthermore, the step (2) comprises: (2.1) Construct a traffic flow uncertainty prediction model based on quantile spatiotemporal network, including the construction of traffic flow data temporal characteristics extraction module, traffic flow data spatial characteristics extraction module, traffic flow data spatiotemporal fusion module and traffic flow uncertainty prediction quantification module based on quantile: (2.1.1) Constructing a traffic flow data temporal feature extraction module: This module is composed of an LSTM network. At time t, the temporal feature extraction module obtains , the different time Splicing to get timing characteristics , (2.1.2) Constructing the spatial feature extraction module of traffic flow data: This module consists of a static feature extraction module and a dynamic feature extraction module. The static feature extraction module is specifically expressed as: , in, represents the static spatial characteristics, represents traffic flow data, , represents the unit diagonal matrix, and Indicates Layer and +1 layer of neural network weight matrix, is a nonlinear activation function, Representation layer normalization, is the adjacency matrix, which is used to represent the road distance and connection relationship between sensor nodes.

[0007] The dynamic feature extraction module is specifically expressed as: , in Represents dynamic spatial characteristics, is the dynamic spatial incidence matrix, , represents traffic flow data, , , , , are learnable weights, The static feature extraction module and the dynamic feature extraction module are integrated to obtain the output of the traffic flow data spatial feature extraction module. , specifically expressed as, ,in represents a fully connected neural network, (2.1.3) Constructing the spatiotemporal fusion module of traffic flow data: The traffic flow data spatiotemporal fusion module is specifically expressed as: , in, , , That is The learnable weight parameters of the head, , , represents the data matrix, Indicates The output of a head, , , , is the corresponding matrix dimension, softmax is the activation function, represents the output weight, concat represents the concatenation function, represents the output of the spatiotemporal fusion module of traffic flow data, represents the matrix transpose, is the set of real numbers, (2.1.4) Traffic flow uncertainty prediction quantification module: In order to quantify the uncertainty of traffic flow, the spatiotemporal fusion module of traffic flow data is converted into quantile prediction results at different quantile levels, which can be specifically expressed as follows: , In the formula, For the The sample in Percentile level The quantile prediction results under Refers to the traffic flow data spatiotemporal fusion module The sample Output results.

[0008] To this end, the goal of traffic flow uncertainty prediction quantification is to minimize the loss, which is expressed as: , in, Indicates The true value of the samples, is the quantile level, is an arbitrarily small positive number, The index aggregation for the dataset, is the index set of quantiles, represents the cardinality function, is the importance of the quantile, Indicated at the quantile level Quantifying model losses for traffic flow prediction under uncertainty; (2.2) Train the traffic flow uncertainty prediction model based on the quantile spatiotemporal network in (2.1), and use the trained model and data to predict traffic flow uncertainty. (2.2.1) Train the traffic flow uncertainty prediction model based on quantile spatiotemporal network in (2.1): The uncertainty prediction model is trained using traffic flow data, and the training objective is to minimize the loss described in (2.1.4). (2.2.2) Use the trained model and data to predict traffic flow uncertainty and realize the prediction of traffic flow uncertainty.

[0009] Furthermore, the step (3) comprises: (3.1) Construct a charging station electric vehicle queuing model that takes into account charging station electricity prices, traffic flow, and road congestion: make For the moment The number of electric vehicles in the charging station at the time, the electric vehicle queuing model conforms to the state space The birth and death process, in which the birth rate The extinction rate Respectively expressed as: , , in, is the number of charging piles in the charging station, is the maximum capacity of the charging station, is the number of electric vehicles in the current charging stations, represents the arrival rate of electric vehicles, The departure rate refers to the average number of electric vehicles leaving the charging station within a fixed period of time. , , is the sensitivity coefficient of electric vehicle users who want to charge to the queue length, is the sensitivity coefficient of the electric car user who wants to leave the charging station with respect to the captain, Indicates the electricity price, Indicates the degree of road congestion. And let the number of electric vehicles in the charging station be The probability of , The corresponding steady-state equation is expressed as: , , (3.2) Using the model constructed in (3.1), the traffic flow uncertainty results are converted into charging station load uncertainty results: Solving the model described in (3.1), we obtain: , , in, , , It means that when the arrival rate is The average number of electric vehicles in the charging station at After calculating the average number of electric vehicles at the charging station in each time period, the average charging load in each time period is calculated, which is expressed as: , in, is the rated charging power of the charging pile, is the load power of the charging station, that is, the uncertainty of the charging load of the charging station. The uncertainty of traffic flow is obtained by calculation. The uncertainty of charging load at the charging station is obtained by this formula.

[0010] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the method for quantifying the uncertainty of charging station load based on spatiotemporal traffic flow prediction is implemented.

[0011] A computer-readable storage medium stores computer instructions, which, when executed by a processor, implement the charging station load uncertainty quantification method based on spatiotemporal traffic flow prediction.

[0012] Compared with the prior art, the charging station load uncertainty quantification method based on spatiotemporal traffic flow prediction provided by the embodiment of the present invention has the following beneficial effects: (1) The traffic flow uncertainty prediction algorithm proposed in this invention constructs a temporal feature extraction module, a spatial feature extraction module and a spatiotemporal fusion module, which can fully exploit the spatiotemporal correlation characteristics of the traffic flow data of the charging station nodes and improve the accuracy of power generation aggregation; (2) Different from the prior art, the present invention constructs a charging station queuing model that takes into account the impact of charging station electricity prices, traffic flow and road congestion factors. Taking into account various realistic factors, it is possible to accurately transform traffic flow uncertainty prediction results into charging station load uncertainty prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a flow chart of a charging station load uncertainty quantification method based on spatiotemporal traffic flow prediction provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0014] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0015] Example: Figure 1 As shown, this embodiment provides a charging station load uncertainty quantification based on quantile spatiotemporal traffic flow prediction, including: (1) Collect traffic flow data from sensor nodes on both sides of the road, including traffic volume data and average vehicle speed data; (2) Construct and train a traffic flow uncertainty prediction model based on quantile spatiotemporal networks, and use this model and traffic flow data to predict traffic flow uncertainty; (3) Construct an electric vehicle queuing model for charging stations that takes into account charging electricity prices, traffic flow, and road congestion, and convert the traffic flow uncertainty prediction results into charging station load uncertainty prediction results.

[0016] The specific steps include: Step 1: Collect traffic flow data from sensor nodes on both sides of the road, including traffic flow data and average vehicle speed data, which is specifically expressed as ,in is the total traffic flow data, express The traffic flow data at that time, N is the number of nodes, T is the time length, H is the number of traffic flow data features of each node, and R represents a real number set.

[0017] Step 2: Build and train a traffic flow uncertainty prediction model based on quantile spatiotemporal networks, and use the model and traffic flow data to predict traffic flow uncertainty; (2.1) Construct a traffic flow uncertainty prediction model based on quantile spatiotemporal network, including the construction of traffic flow data temporal characteristics extraction module, traffic flow data spatial characteristics extraction module, traffic flow data spatiotemporal fusion module and traffic flow uncertainty prediction quantification module based on quantile: (2.1.1) Constructing a traffic flow data temporal feature extraction module: This module is composed of an LSTM network. At time t, the temporal feature extraction module obtains , the different time Splicing to get timing characteristics , (2.1.2) Constructing the spatial feature extraction module of traffic flow data: This module consists of a static feature extraction module and a dynamic feature extraction module. The static feature extraction module is specifically expressed as: , in, represents the static spatial characteristics, represents traffic flow data, , represents the unit diagonal matrix, and Indicates Layer and +1 layer of neural network weight matrix, is a nonlinear activation function, Representation layer normalization, is the adjacency matrix, which is used to represent the road distance and connection relationship between sensor nodes.

[0018] The dynamic feature extraction module is specifically expressed as: , in Represents dynamic spatial characteristics, is the dynamic spatial incidence matrix, , represents traffic flow data, , , , , are learnable weights, The static feature extraction module and the dynamic feature extraction module are integrated to obtain the output of the traffic flow data spatial feature extraction module. , specifically expressed as, ,in represents a fully connected neural network, (2.1.3) Constructing the spatiotemporal fusion module of traffic flow data: The traffic flow data spatiotemporal fusion module is specifically expressed as: , in, , , That is The learnable weight parameters of the head, , , represents the data matrix, Indicates The output of a head, , , , is the corresponding matrix dimension, softmax is the activation function, represents the output weight, concat represents the concatenation function, represents the output of the spatiotemporal fusion module of traffic flow data, represents the matrix transpose, is the set of real numbers, (2.1.4) Traffic flow uncertainty prediction quantification module: In order to quantify the uncertainty of traffic flow, the spatiotemporal fusion module of traffic flow data is converted into quantile prediction results at different quantile levels, which can be specifically expressed as follows: , In the formula, For the The sample in Percentile level The quantile prediction results under Refers to the traffic flow data spatiotemporal fusion module The sample Output results.

[0019] To this end, the goal of traffic flow uncertainty prediction quantification is to minimize the loss, which is expressed as: , , , in, Indicates The true value of the samples, is the quantile level, is an arbitrarily small positive number, The index aggregation for the dataset, is the index set of quantiles, represents the cardinality function, is the importance of the quantile, Indicated at the quantile level Quantifying model losses for traffic flow prediction under uncertainty; (2.2) Train the traffic flow uncertainty prediction model based on the quantile spatiotemporal network in (2.1), and use the trained model and data to predict traffic flow uncertainty. (2.2.1) Train the traffic flow uncertainty prediction model based on quantile spatiotemporal network in (2.1): The uncertainty prediction model is trained using traffic flow data, and the training objective is to minimize the loss described in (2.1.4). (2.2.2) Use the trained model and data to predict traffic flow uncertainty and realize the prediction of traffic flow uncertainty.

[0020] Step 3: Construct an electric vehicle queuing model for charging stations that takes into account charging electricity prices, traffic flow, and road congestion, and convert the traffic flow uncertainty prediction results into charging station load uncertainty prediction results. (3.1) Construct a charging station electric vehicle queuing model that takes into account charging station electricity prices, traffic flow, and road congestion: make For the moment The number of electric vehicles in the charging station at the time, the electric vehicle queuing model conforms to the state space The birth and death process, in which the birth rate The extinction rate Respectively expressed as: , , in, is the number of charging piles in the charging station, is the maximum capacity of the charging station, is the number of electric vehicles in the current charging stations, represents the arrival rate of electric vehicles, The departure rate refers to the average number of electric vehicles leaving the charging station within a fixed period of time. , , is the sensitivity coefficient of electric vehicle users who want to charge to the queue length, is the sensitivity coefficient of the electric car user who wants to leave the charging station with respect to the captain, Indicates the electricity price, Indicates the degree of road congestion. And let the number of electric vehicles in the charging station be The probability of , The corresponding steady-state equation is expressed as: , , (3.2) Using the model constructed in (3.1), the traffic flow uncertainty results are converted into charging station load uncertainty results: Solving the model described in (3.1), we obtain: , , in, , , It means that when the arrival rate is The average number of electric vehicles in the charging station at After calculating the average number of electric vehicles at the charging station in each time period, the average charging load in each time period is calculated, which is expressed as: , in, is the rated charging power of the charging pile, is the load power of the charging station, that is, the uncertainty of the charging load of the charging station. The uncertainty of traffic flow is obtained by calculation. The uncertainty of charging load at the charging station is obtained by this formula.

[0021] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A charging station load uncertainty quantification method based on spatiotemporal traffic flow prediction, characterized in that: The following steps are involved: (1) Collect traffic flow data from sensor nodes on both sides of the road, including traffic volume data and average vehicle speed data; (2) Construct and train a traffic flow uncertainty prediction model based on quantile spatiotemporal networks, and use this model and traffic flow data to predict traffic flow uncertainty; (3) Construct an electric vehicle queuing model for charging stations that takes into account charging electricity prices, traffic flow, and road congestion, and convert the traffic flow uncertainty prediction results into charging station load uncertainty prediction results.

2. The charging station load uncertainty quantification method based on spatiotemporal traffic flow prediction according to claim 1 is characterized in that: Step (1) is as follows: Traffic flow data is collected from sensor nodes on both sides of the road, including traffic flow data and average vehicle speed data, which can be expressed as ,in is the total traffic flow data, express The traffic flow data at that time, N is the number of nodes, T is the time length, H is the number of traffic flow data features of each node, and R represents a real number set.

3. The charging station load uncertainty quantification method based on spatiotemporal traffic flow prediction according to claim 1 is characterized in that: Step (2) is as follows: (2.1) Construct a traffic flow uncertainty prediction model based on quantile spatiotemporal network, including the construction of traffic flow data temporal characteristics extraction module, traffic flow data spatial characteristics extraction module, traffic flow data spatiotemporal fusion module and traffic flow uncertainty prediction quantification module based on quantile: (2.1.1) Constructing a traffic flow data temporal feature extraction module: This module is composed of an LSTM network. The time series feature extraction module obtains , the different time Splicing to get timing characteristics , (2.1.2) Constructing the spatial feature extraction module of traffic flow data: This module consists of a static feature extraction module and a dynamic feature extraction module. The static feature extraction module is specifically expressed as: ， in, Represents the static spatial characteristics, represents traffic flow data, , represents the unit diagonal matrix, and Indicates Layer and +1 layer of neural network weight matrix, is a nonlinear activation function, Representation layer normalization, is the adjacency matrix, which is used to represent the road distance and connection relationship between sensor nodes. The dynamic feature extraction module is specifically expressed as: ， in Represents dynamic spatial characteristics, is the dynamic spatial incidence matrix, , represents traffic flow data, , , , , are learnable weights, The static feature extraction module and the dynamic feature extraction module are integrated to obtain the output of the traffic flow data spatial feature extraction module. , specifically expressed as, ,in represents a fully connected neural network, (2.1.3) Constructing the spatiotemporal fusion module of traffic flow data: The traffic flow data spatiotemporal fusion module is specifically expressed as: ， in, , , That is The learnable weight parameters of the head, , , represents the data matrix, Indicates The output of a head, , , , is the corresponding matrix dimension, softmax is the activation function, represents the output weight, concat represents the concatenation function, represents the output of the spatiotemporal fusion module of traffic flow data, represents the matrix transpose, is the set of real numbers, (2.1.4) Traffic flow uncertainty prediction quantification module: In order to quantify the uncertainty of traffic flow, the spatiotemporal fusion module of traffic flow data is converted into quantile prediction results at different quantile levels, which can be specifically expressed as follows: ， In the formula, For the The sample in Percentile level The quantile prediction results under Refers to the traffic flow data spatiotemporal fusion module The sample Output results, The goal of traffic flow uncertainty prediction quantification is to minimize the loss, which is expressed as: ， . in, Indicates The true value of the samples, is the quantile level, is an arbitrarily small positive number, The index aggregation for the dataset, is the index set of quantiles, represents the cardinality function, is the importance of the quantile, Indicated at the quantile level Quantifying model losses for traffic flow prediction under uncertainty, (2.2) Train the traffic flow uncertainty prediction model based on the quantile spatiotemporal network in (2.1), and use the trained model and data to predict traffic flow uncertainty. (2.2.1) Train the traffic flow uncertainty prediction model based on quantile spatiotemporal network in (2.1): The uncertainty prediction model is trained using traffic flow data, and the training objective is to minimize the loss described in (2.1.4). (2.2.2) Use the trained model and data to predict traffic flow uncertainty and realize the prediction of traffic flow uncertainty.

4. The charging station load uncertainty quantification method based on spatiotemporal traffic flow prediction according to claim 1 is characterized in that: Step (3) is as follows: (3.1) Construct a charging station electric vehicle queuing model that takes into account charging station electricity prices, traffic flow, and road congestion: make For the moment The number of electric vehicles in the charging station at the time, the electric vehicle queuing model conforms to the state space The birth and death process, in which the birth rate The extinction rate Respectively expressed as: in, is the number of charging piles in the charging station, is the maximum capacity of the charging station, is the number of electric vehicles in the current charging stations, represents the arrival rate of electric vehicles, The departure rate refers to the average number of electric vehicles leaving the charging station within a fixed period of time. , , is the sensitivity coefficient of electric vehicle users who want to charge to the queue length, is the sensitivity coefficient of the electric car user who wants to leave the charging station with respect to the captain, Indicates the electricity price, Indicates the degree of road congestion. And let the number of electric vehicles in the charging station be The probability of , The corresponding steady-state equation is expressed as: ， (3.2) Using the model constructed in (3.1), the traffic flow uncertainty results are converted into charging station load uncertainty results: Solving the model described in (3.1), we obtain: ， ， in, , , It means that when the arrival rate is The average number of electric vehicles in the charging station at After calculating the average number of electric vehicles at the charging station in each time period, the average charging load in each time period is calculated, which is expressed as: ， in, is the rated charging power of the charging pile, is the load power of the charging station.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for quantifying uncertainty of charging station load based on spatiotemporal traffic flow prediction as described in any one of claims 1 to 4 above is implemented.

6. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by a processor, a charging station load uncertainty quantification method based on spatiotemporal traffic flow prediction as described in any one of claims 1 to 4 is implemented.

Images