A dynamic estimation method for energy consumption and carbon emission of electric hybrid traffic flow based on space-time diagram network

Abstract

The application discloses a kind of dynamic estimation method of energy consumption carbon emission of electric hybrid traffic flow based on space-time graph network, which comprises establishing the directed graph structure of selected road network, constructing the dynamic adjacency matrix based on real vehicle trip data, obtaining the dynamic spatial correlation between road segment nodes under the directed graph structure;Based on the real road, traffic flow, running speed, temperature, relative humidity collected, the input-output sample set under the directed graph structure of road network is established;A flow-running-energy consumption / carbon emission response relationship model is constructed, the model is trained using the input-output sample set, and the optimal model obtained by training is used to realize dynamic estimation of energy consumption carbon emission under different traffic flow inputs.The application creates a standard road network graph framework under the dynamic estimation method of energy consumption, carbon emission of different types of vehicles at different space-time positions, which has good replicability in different regions.

Description

Technical Field

[0001] This invention relates to the field of low-carbon assessment technology for urban road traffic, specifically to a dynamic estimation method for energy consumption and carbon emissions of electrified hybrid traffic flow based on spatiotemporal graph networks. Background Technology

[0002] With the vigorous promotion of new energy vehicles and intelligent transportation technologies, vehicle electrification and intelligent connected hybrid transportation are gradually forming a new road traffic operation mode. For a considerable period, the future road traffic environment will be characterized by a mix of vehicles using multiple energy sources, including gasoline, electric, and hybrid vehicles. The spatiotemporal estimation of energy consumption and carbon emissions from mixed-vehicle traffic on dynamic road networks has become a critical challenge for reducing carbon emissions in this new hybrid transportation mode. Current developments in artificial intelligence technologies, such as deep learning, can provide new technical approaches for the dynamic estimation of energy consumption and carbon emissions from electrified mixed-vehicle flows on dynamic road networks.

[0003] Energy consumption and carbon emissions from vehicles on the road are comprehensively influenced by factors such as vehicle technology (brand, year, type, fuel type, engine, after-treatment system, etc.) and traffic conditions (road conditions, driving status, human habits, etc.), exhibiting significant dynamic time-varying characteristics. Estimating energy consumption and carbon emissions from traffic flow on real road networks requires inputs including vehicle technology information, traffic status data, road network information, and weather data. By acquiring the dynamic activity level of travel, an estimation model at the corresponding scale can be established. Estimating energy consumption and carbon emissions on dynamic road networks involves spatiotemporal complexity due to various vehicle types, time-varying operations, and road spatial distribution. Current research often focuses on single vehicle types or uses segment-by-segment estimation methods, resulting in estimations that often ignore spatial correlations within the road network, and the computational process suffers from poor system reproducibility. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic estimation method for energy consumption and carbon emissions of electrified hybrid vehicle flows based on spatiotemporal graph networks, which can effectively improve the system efficiency of estimation and has good reproducibility for different regions.

[0005] This invention adopts the following technical solution: a method for dynamic estimation of energy consumption and carbon emissions of electrified hybrid vehicle flow based on spatiotemporal graph networks, comprising the following steps:

[0006] S1. Establish a directed graph structure for the selected road network, construct a dynamic adjacency matrix based on real vehicle travel data, and obtain the dynamic spatial association between road segment nodes under the directed graph structure.

[0007] S2. Based on the collected real road data, traffic flow, operating speed, temperature, and relative humidity, establish an input-output sample set under a directed graph structure of the road network.

[0008] S3. Construct a spatiotemporal map feature extraction module that associates multiple elements such as road, vehicle, operation, and weather. This module includes sequentially connected convolutional layers and associated gating units.

[0009] S4. Construct a traffic flow-operation-energy consumption / carbon emission response relationship model. Divide the input-output sample set into a training set and a test set according to a set ratio. Use the training set to train the model. Input the input sample set from the test set into the trained traffic flow-operation-energy consumption / carbon emission response relationship model to obtain the dynamic estimation results of energy consumption and carbon emissions.

[0010] Furthermore, in step S1, the dynamic spatial relationships between road segment nodes are obtained, including the following:

[0011] By defining road segments as nodes and the lines connecting them as edges, the road network can be represented as a directed graph structure G. V represents the set of road segment nodes. , Let N represent the i-th road segment node, and N represent the total number of road segment nodes. Describe the set of edges. This represents an adjacency matrix, used to display the dynamic spatial relationships between road segment nodes. , The adjacency matrix represents the relationship between the i-th road segment node and the j-th road segment node, where j=1,2,…,N.

[0012] The resolution of the adjacency matrix is ​​the same as that of real vehicle travel data.

[0013] Based on the connectivity between road segments and the bidirectional traffic flow direction, a basic adjacency matrix of a directed graph structure is constructed. The specific expression is:

[0014]

[0015] Among them, V i V represents the i-th road segment node. j This represents the j-th road segment node.

[0016] Let T represent the time series set. , Let C represent the t-th time series and C represent the total number of time series; obtain the traffic flow of different vehicle types at each road segment node in the road network. , , Indicates the first Traffic flow of vehicle type type N at time n on the Nth road segment node.

[0017] According to the limit space range, the first Calculate the reciprocal of the difference in traffic flow at each moment. Time-of-movement traffic flow similarity weight matrix The specific expression is:

[0018]

[0019] in, Indicates the first The difference in traffic flow between the i-th road segment node and the j-th road segment node for the i-th vehicle type at time t. Indicates the first The traffic flow of vehicle type 'type' at time 'i' at the i-th road segment node. Indicates the first The traffic flow of vehicle type j at time j on the road segment node. This represents the minimum distance between two road segment nodes. V represents i With V j The distance between them.

[0020] Will As a dynamic enhancement factor of the adjacency matrix .

[0021] The adjacency matrix is ​​obtained by fusing the basic adjacency matrix with the dynamic enhancement factor of the adjacency matrix. The specific expression is as follows:

[0022] .

[0023] Furthermore, in step S2, establishing the input-output sample set includes the following:

[0024] Based on collected real-world road data, traffic flow, operating speed, temperature, and relative humidity, a standard dataset Data is constructed using a directed graph structure, represented as follows:

[0025]

[0026] in, Indicates the length of the road segment. This indicates the traffic flow for different vehicle types. Indicates the speed of the road segment. Indicates temperature, This indicates relative humidity.

[0027] The standard dataset Data is processed using data augmentation at standard spatiotemporal resolution to obtain the input sample set, including:

[0028] Outliers in real roads, traffic flow, and operating speed are removed by defining limit values. Values ​​that deviate from the overall average by more than three times are considered outliers. Missing values ​​in real roads, traffic flow, and operating speed are supplemented by using the nearest neighbor mean.

[0029] The inverse distance-weighted difference is used to perform spatiotemporal processing on air temperature and relative humidity, with the following formula:

[0030]

[0031]

[0032] in, Indicates the first The temperature at the i-th road segment node at time i, where IDW represents inverse distance weighted interpolation. This represents the S-th temperature. Indicates the first The relative humidity of the i-th road segment node at time i. This represents the S-th relative humidity.

[0033] The energy consumption of different vehicle types under a directed graph structure is estimated using a theoretical estimation model, resulting in a standard energy consumption dataset. The specific formula is as follows:

[0034]

[0035] in, Indicates the first Energy consumption of vehicle type type i at time i on road segment node; This represents the length of the i-th segment node; This represents the energy consumption factor of vehicle type 'type' under the conditions of road segment operating speed, temperature, and relative humidity.

[0036] Carbon emissions from gasoline-powered vehicles are estimated using a speed-based emission factor model within a directed graph structure, resulting in a standard carbon emission dataset. The specific formula is as follows:

[0037]

[0038] in, Indicates the first The carbon emissions of vehicle type i at time i on road segment node i. This represents the carbon emission factor of vehicle type 'type' under operating speed, temperature, and relative humidity.

[0039] Based on the energy consumption standard dataset and the carbon emission standard dataset, the output sample set is obtained. , .

[0040] Furthermore, in step S4, the dynamic estimation results of energy consumption and carbon emissions include the following:

[0041] The traffic flow-operation-energy consumption / carbon emission response model includes a spatiotemporal graph feature extraction module, a spatiotemporal attention graph convolution module, and an LSTM module that are sequentially connected to multiple factors related to roads, vehicles, operation, and weather.

[0042] The input sample set is composed of a multidimensional spatiotemporal tensor. Then, the features are input into the trained traffic flow-operation-energy consumption / carbon emission response model according to different feature dimensions. After passing through the spatiotemporal graph feature extraction module that associates multiple factors such as road-vehicle-operation-weather, low-dimensional features are extracted using a convolutional layer with a kernel size of 1*1. These features are then passed through an association gating unit to obtain the corresponding association feature information. The specific formula is as follows:

[0043]

[0044]

[0045]

[0046] in, This represents the correlation information between traffic flow and road segment operating speed for the vehicle type of the specified type. This represents the correlation between traffic flow and temperature for the vehicle type 'type'. This represents the correlation information between traffic flow and relative humidity for the vehicle type 'type'. This represents a low-dimensional feature of the traffic flow for the 'type' vehicle type. Low-dimensional features representing the operating speed of road segments, Representing the low-dimensional characteristics of temperature, This represents a low-dimensional characteristic of relative humidity.

[0047] Will , , , , Weighted fusion yields graph feature vectors representing the relationships between multiple factors at different times. The specific formula is as follows:

[0048] .

[0049] in, represents the low-dimensional feature of road segment length, and F represents weighted fusion.

[0050] Graph feature vectors that link multiple elements After being fused with the dynamic adjacency matrix, the graph convolution operation is performed by the spatiotemporal attention graph convolution module to obtain the feature vector at each time step. This feature vector is then passed through the LSTM module to obtain the time-related features, thus completing the dynamic estimation of energy consumption and carbon emissions.

[0051] Furthermore, the present invention also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the dynamic estimation method for energy consumption and carbon emissions of electrified hybrid vehicle flow based on spatiotemporal graph networks.

[0052] Furthermore, the present invention also proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the aforementioned method for dynamic estimation of energy consumption and carbon emissions of electrified hybrid vehicle flows based on spatiotemporal graph networks.

[0053] Compared with the prior art, the present invention, employing the above technical solution, has the following technical effects:

[0054] This invention enables synchronous dynamic estimation of energy consumption and carbon emissions for different types of vehicles at different times and locations within a standard map framework. It solves the problem of object estimation under the dynamic and complex structure of road networks, effectively improves the system efficiency of estimation, and has good reproducibility for different regions. Attached Figure Description

[0055] Figure 1 This is a flowchart illustrating the overall implementation of the present invention.

[0056] Figure 2 This is a schematic diagram of the traffic flow-operation-energy consumption / carbon emission response model of the present invention.

[0057] Figure 3 This is a graph showing the carbon emission estimation results of an embodiment of the present invention. Detailed Implementation

[0058] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0059] To achieve the above objectives, this invention proposes a dynamic estimation method for energy consumption and carbon emissions of electrified hybrid vehicle flows based on spatiotemporal graph networks, such as... Figure 1 As shown, the specific steps are as follows:

[0060] S1. Establish a directed graph structure for the selected road network, construct a dynamic adjacency matrix based on real vehicle travel data, and obtain the dynamic spatial relationships between road segment nodes under the directed graph structure. Specific content includes:

[0061] By defining road segments as nodes and the lines connecting them as edges, the road network can be represented as a directed graph structure G. V represents the set of road segment nodes. , Let N represent the i-th road segment node, and N represent the total number of road segment nodes. Describe the set of edges. This represents the adjacency matrix, used to represent topological relationships and describe the spatial associations between nodes. , The adjacency matrix represents the relationship between the i-th road segment node and the j-th road segment node, where j=1,2,…,N.

[0062] In order to characterize the spatial correlation of traffic flow at different times, the dynamic adjacency matrix is ​​estimated using real traffic flow at the same frequency time period. The time resolution is the same as the resolution of the collected real vehicle travel data. That is, the real road network traffic flow data is obtained with a time resolution of 1 hour, and the time resolution of the dynamic adjacency matrix estimation is also 1 hour.

[0063] Based on the connectivity between road segments and the bidirectional traffic flow direction, a basic adjacency matrix of a directed graph structure is constructed. The specific expression is:

[0064]

[0065] Among them, V i V represents the i-th road segment node. j This represents the j-th road segment node.

[0066] Represents a node There are upstream and downstream connections between them. Represents a node There is no upstream or downstream connection between them.

[0067] Let T represent the time series set. , Let C represent the t-th time series and C represent the total number of time series; obtain the traffic flow of different vehicle types at each road segment node in the road network. , , Indicates the first Traffic flow of vehicle type type N at time n on the Nth road segment node.

[0068] According to the limit space range, the first Calculate the reciprocal of the difference in traffic flow at each moment. Time-of-movement traffic flow similarity weight matrix The specific expression is:

[0069]

[0070] in, Indicates the first The difference in traffic flow between the i-th road segment node and the j-th road segment node for the i-th vehicle type at time t; Indicates the first Traffic flow of vehicle type 'type' at the i-th road segment node at time 1; Indicates the first Traffic flow of vehicle type j at time j on road segment node; This represents the minimum distance between two road segment nodes. For example, if the minimum distance is set to 3 kilometers, it means that only the relationship between adjacent road segment nodes within a 3-kilometer range is considered. V represents i With V j The distance between them.

[0071] When the distance between two road segment nodes is within the limit, the reciprocal of the difference in traffic flow between the two road segments is taken as the dynamic enhancement factor to enhance the correlation between road segment nodes. When it is not within the limit, the dynamic enhancement factor is 0.

[0072] Will As a dynamic enhancement factor of the adjacency matrix .

[0073] The adjacency matrix is ​​obtained by fusing the basic adjacency matrix with the dynamic enhancement factor of the adjacency matrix. The specific expression is as follows:

[0074] .

[0075] S2. Based on collected real-world road data, traffic flow, speed, temperature, and relative humidity, establish an input-output sample set for the road network under a directed graph structure. Specific details are as follows:

[0076] Based on collected real-world road data, traffic flow, operating speed, temperature, and relative humidity, a standard dataset named Data is constructed using a directed graph structure to achieve spatiotemporal correlation between multiple data elements, represented as follows:

[0077]

[0078] in, Indicates the length of the road segment. This indicates the traffic flow for different vehicle types. Indicates the speed of the road segment. Indicates temperature, This indicates relative humidity.

[0079] The standard dataset Data is processed using data augmentation at standard spatiotemporal resolution to obtain the input sample set, including:

[0080] Outliers in real roads, traffic flow, and operating speed are removed by defining limit values. Values ​​that deviate from the overall average by more than three times are considered outliers. Missing values ​​in real roads, traffic flow, and operating speed are supplemented by using the nearest neighbor mean.

[0081] Inverse distance weighted difference is used to perform spatiotemporal processing on temperature and relative humidity, completing the spatiotemporal data estimation from monitoring stations to the directed graph structure of the urban road network. This yields temperature and relative humidity datasets within the directed graph structure of the urban road network, achieving the same spatiotemporal resolution for weather data as for traffic flow and speed data. The specific formula is as follows:

[0082]

[0083]

[0084] in, Indicates the first The temperature at the i-th road segment node at time i, where IDW represents inverse distance weighted interpolation. This represents the S-th temperature. Indicates the first The relative humidity of the i-th road segment node at time i. This represents the S-th relative humidity.

[0085] The energy consumption of different vehicle types under a directed graph structure is estimated using a theoretical estimation model, resulting in a standard energy consumption dataset. The specific formula is as follows:

[0086]

[0087] in, Indicates the first Energy consumption of vehicle type type i at time i on road segment node; This represents the length of the i-th segment node; This represents the energy consumption factor of vehicle type 'type' under the conditions of road segment operating speed, temperature, and relative humidity.

[0088] Carbon emissions from gasoline-powered vehicles are estimated using a speed-based emission factor model within a directed graph structure, resulting in a standard carbon emission dataset. The specific formula is as follows:

[0089]

[0090] in, Indicates the first The carbon emissions of vehicle type 'type' at time 'i' at road segment node, in kg; This represents the carbon emission factor of vehicle type 'type' under operating speed, temperature, and relative humidity, expressed in kg / km.

[0091] Based on the energy consumption standard dataset and the carbon emission standard dataset, the output sample set is obtained. , .

[0092] S3. Construct a spatiotemporal map feature extraction module that associates multiple elements such as road, vehicle, operation, and weather. This module includes sequentially connected convolutional layers and associated gating units.

[0093] S4. Construct a vehicle flow-operation-energy consumption / carbon emission response model. Divide the input-output sample set into a training set and a test set according to a set ratio. Use the training set to train the model. Input the input sample set from the test set into the trained vehicle flow-operation-energy consumption / carbon emission response model to obtain dynamic estimation results of energy consumption and carbon emissions. Figure 2 As shown, the specific content is as follows:

[0094] The traffic flow-operation-energy consumption / carbon emission response model includes a spatiotemporal graph feature extraction module, a spatiotemporal attention graph convolution module, and an LSTM module that are sequentially connected to multiple factors related to roads, vehicles, operation, and weather.

[0095] The input sample set is composed of a multidimensional spatiotemporal tensor. Then, the features are input into the trained traffic flow-operation-energy consumption / carbon emission response model according to different feature dimensions. After passing through the spatiotemporal graph feature extraction module that associates multiple factors such as road-vehicle-operation-weather, low-dimensional features are extracted using a convolutional layer with a kernel size of 1*1. These features are then passed through an association gating unit to obtain the corresponding association feature information. The specific formula is as follows:

[0096]

[0097]

[0098]

[0099] in, This represents the correlation information between traffic flow and road segment operating speed for the vehicle type of the specified type. This represents the correlation between traffic flow and temperature for the vehicle type 'type'. This represents the correlation information between traffic flow and relative humidity for the vehicle type 'type'. This represents a low-dimensional feature of the traffic flow for the 'type' vehicle type. Low-dimensional features representing the operating speed of a road segment. Representing the low-dimensional characteristics of temperature, This represents a low-dimensional characteristic of relative humidity.

[0100] Will , , , , Weighted fusion yields graph feature vectors representing the relationships between multiple factors at different times. The specific formula is as follows:

[0101] .

[0102] in, represents the low-dimensional feature of road segment length, and F represents weighted fusion.

[0103] Graph feature vectors that link multiple elements After being fused with the dynamic adjacency matrix, the graph convolution operation is performed by the spatiotemporal attention graph convolution module to obtain the feature vector at each time step. This feature vector is then passed through the LSTM module to obtain the time-related features, thus completing the dynamic estimation of energy consumption and carbon emissions.

[0104] Example:

[0105] The road network of a core urban area was selected, covering an area of ​​approximately 2,000 square kilometers and containing 3,239 road segments. Data was collected on actual roads with segment location information, names, and lengths; hourly traffic flow data for each road segment over a month; and hourly temperature and relative humidity data from 10 meteorological monitoring stations. C is set to 744, meaning there are 744 time points.

[0106] The following is a calculation result at a certain moment in the carbon emission estimation of gasoline-powered passenger cars:

[0107] .

[0108] Vehicle classification is based on national classification standards. In this embodiment, vehicles are categorized by fuel type (gasoline, pure electric, hybrid) and by vehicle type (small passenger car, large passenger car, small truck, large truck, bus, taxi). Therefore, there are 18 vehicle types in total. However, not every vehicle type requires energy consumption and carbon emission calculations. Some vehicle types do not require carbon emission estimation, only energy consumption estimation; for example, the six types of gasoline vehicles.

[0109] Based on the collected data, an input-output sample set was constructed, in which the input-output sample set of the first 30 days was used to train the model and obtain the optimal model.

[0110] Using the optimal model and the data from the last 24 hours of the previous day as the test sample, carbon emissions were estimated under a 24-hour input of fuel-powered passenger car traffic flow, yielding 24-hour carbon emission estimation results under the standard graph structure. A visualization of the carbon emission results obtained under the 18:00 fuel-powered passenger car traffic flow input is attached. Figure 3 .from Figure 3 From (a), we can see the distribution of the 3239 road segment nodes on the selected example road network. Figure 3 As can be seen in (b), the carbon emission results of fuel-powered passenger cars at 18 hours were obtained using the method proposed in this invention. Therefore, it can be seen that the method proposed in this invention can estimate the energy consumption or carbon emissions of different types of vehicles. Obviously, the estimation results under the standard road network map structure were finally obtained.

[0111] Furthermore, the vehicle flow-operation-energy consumption / carbon emission response model proposed in this invention was evaluated using MAE, RMSE, MAPE, and R... 2 As evaluation metrics for the model, smaller MAE, RMSE, MAPE, and R² values ​​close to 1 are considered. 2 The values ​​all reflect more accurate and reliable estimation results.

[0112] Finally, the estimation results obtained by the model proposed in this invention were compared with those of a general LSTM network model and a general neural network model that removes the spatiotemporal graph feature extraction module and the spatiotemporal attention graph convolution module. The comparison results are shown in Table 1.

[0113] Table 1. Results of dynamic estimation of energy consumption and carbon emissions of traffic flow using different models.

[0114]

[0115] The comparison results show that, compared with the comparative model, the model described in this technology proposes a standard road network graph structure representation and spatiotemporal graph feature extraction under the standard graph structure. Combined with the spatiotemporal attention graph convolution module, it can achieve better estimation results.

[0116] This invention also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. It should be noted that when the processor executes the computer program, it corresponds to the specific steps of the method provided in this invention, possessing the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in this invention.

[0117] This invention also proposes a computer-readable storage medium storing a computer program. It should be noted that when the computer program is executed by a processor, it corresponds to the specific steps of the method provided in this invention, possessing the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in this invention.

[0118] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A spatiotemporal graph network-based dynamic estimation method for energy consumption and carbon emission of electrified mixed traffic flow, characterized in that, The method comprises the steps of: S1, establishing a directed graph structure of a selected road network, constructing a dynamic adjacency matrix based on real vehicle travel data, and obtaining dynamic spatial correlation between road segment nodes under the directed graph structure; S2, based on the collected real road, traffic flow, running speed, air temperature and relative humidity, an input-output sample set under the directed graph structure of the road network is established; S3, a space-time graph feature extraction module related to road-vehicle-operation-weather multi-element is constructed, which includes convolution layers and correlation gate units connected in turn; S4, a vehicle flow-operation-energy consumption / carbon emission response relationship model is constructed, the input-output sample set is divided into a training set and a test set according to a set proportion, the training set is used to train the model, and the input sample set in the test set is input into the trained vehicle flow-operation-energy consumption / carbon emission response relationship model to obtain dynamic estimation results of energy consumption and carbon emission; Specifically: The vehicle flow-operation-energy consumption / carbon emission response relationship model includes a space-time graph feature extraction module related to road-vehicle-operation-weather multi-element, a space-time attention graph convolution module and an LSTM module connected in turn; The input sample set is composed into a multi-dimensional space-time tensor X ∈ R N,T After that, input into the trained vehicle flow-operation-energy consumption / carbon emission response relationship model respectively according to different feature dimensions, through the space-time graph feature extraction module related to road-vehicle-operation-weather multiple factors, low-dimensional features are extracted by using a convolution layer with a convolution kernel size of 1*1, the features pass through the associated gate unit to obtain the corresponding associated feature information, and the specific formula is: wherein, represents the correlation characteristic information of the traffic volume of the type vehicle type and the road segment running speed, represents the correlation characteristic information of the traffic volume of the type vehicle type and the air temperature, represents the correlation characteristic information of the traffic volume of the type vehicle type and the relative humidity, represents the low-dimensional feature of the traffic volume of the type vehicle type, D Speed represents the low-dimensional feature of the road segment running speed, D AirTem represents the low-dimensional feature of the air temperature, D Rh represents the low-dimensional feature of the relative humidity; D Road 、 Weighted fusion, get different time of multi-element associated graph feature vector X', the specific formula is: where D Road represents the low-dimensional features of road segment length, and F represents the weighted fusion. After the multi-element related graph feature vector X' is fused with the dynamic adjacency matrix, the space-time attention graph convolution module is used for graph convolution operation to obtain a feature vector at each time, and the feature vector is input into the LSTM module to obtain a time correlation feature, thereby completing dynamic estimation of energy consumption and carbon emission.

2. The spatiotemporal graph network-based dynamic estimation method of electrified mixed traffic flow energy consumption and carbon emission according to claim 1, wherein, In step S1, the dynamic spatial correlation between road segment nodes includes the following contents: Set the road section as a node, the connection between the road sections as an edge, and represent the road network as a directed graph structure G, G=(V, E, A), V represents the set of road section nodes, V={V1, V2, …, V i ,...,V N}, V i represents the i-th road section node, N represents the total number of road section nodes, E represents the set of edges, A represents the adjacency matrix, and is used to show the dynamic spatial correlation between the road section nodes, A ij represents the adjacency matrix of the correlation between the i-th road section node and the j-th road section node, j=1, 2, …, N. The resolution of the adjacency matrix is the same as that of the real vehicle travel data; Wherein, according to the connection between the road section, combined with the road two-way traffic flow direction, the construction of directed graph structure basis adjacency matrix A Base ∈R N×N , the specific expression is: wherein V i represents the ith road segment node, V j represents the jth road segment node; Let T represent a set of time series, T = {T1, T2,..., T t ,...,T C}, T t represents the tth time series, and C represents the total number of time series; the traffic flow f type , of different vehicle types of each road segment node in the road network is obtained, where f t represents the traffic flow of the type vehicle type at the Nth road segment node at the Tth time point. According to the Tth value within the limit space t Calculate the reciprocal of the traffic flow difference at time T. t Time-of-movement traffic flow similarity weight matrix The specific expression is: wherein, denotes the number of vehicles of type T t at time t between the i-th and j-th road segment nodes, denotes the number of vehicles of type T t at time t at the i-th road segment node, denotes the number of vehicles of type T t at time t at the j-th road segment node, 限 denotes the minimum limit of the distance between two road segment nodes, |D(V i , V j )| denotes the distance between V i and V j . will be described below. as the dynamic enhancement factor A' of the adjacency matrix; The basic adjacency matrix is fused with the dynamic enhancement factor of the adjacency matrix to obtain the adjacency matrix, and the specific expression is: A = A Base • A'.

3. The spatiotemporal graph network-based dynamic estimation method of electrified mixed traffic flow energy consumption and carbon emission according to claim 1, wherein, In step S2, the input-output sample set is established, including the following contents: Based on the collected real road, traffic flow, running speed, air temperature and relative humidity, a standard data set Data under the directed graph structure is constructed, expressed as: Data = { Road E R N , Traffic type E R N×T , Speed E R N×T , AirTem E R N×T , Rh E R N×T} wherein, Road represents the length of road segment, Traffic type represents the traffic volume of different vehicle types, Speed represents the running speed of road segment, AirTem represents the air temperature, Rh represents the relative humidity, N represents the total number of road segment nodes, and T represents the set of time series. The standard data set Data is processed by using data enhancement with a standard space-time resolution to obtain an input sample set, including: The real road, traffic flow and running speed are subjected to outlier rejection by using a limit value definition method, and values deviating from the overall average value by more than 3 times are regarded as outliers; The real road, traffic flow and running speed are supplemented with missing values by using a nearest neighbor mean value; The air temperature and relative humidity are subjected to space-time processing by using an inverse distance weighted interpolation, and the specific formula is: AirTem(V i ,T t ) = IDW(AirTem1, AirTem2,..., AirTem S ) Rh(V i ,T t ) = IDW(Rh1, Rh2,..., Rh S ) wherein, AirTem(V i , T t ) represents the air temperature of the ith road segment node at the T t th moment, IDW represents the inverse distance weighted interpolation, AirTem S represents the Sth air temperature, Rh(V i , T t ) represents the relative humidity of the ith road segment node at the T t th moment, Rh S represents the Sth relative humidity; The energy consumption of different vehicle types under the directed graph structure is estimated by using a theoretical estimation model to obtain an energy consumption standard data set, and the specific formula is: wherein, denotes the energy consumption of the type-th vehicle type at the i-th road segment node at the T-th time instant; t denotes the energy consumption of the type-th vehicle type at the i-th road segment node at the T-th time instant; denotes the length of the i-th road segment node; En type,speed,AirTem,Rh denotes the energy consumption factor of the type-th vehicle type at the road segment running speed, air temperature, and relative humidity. The carbon emission of the fuel vehicle under the directed graph structure is estimated by using a speed-based emission factor model to obtain a carbon emission standard data set, and the specific formula is: wherein, denotes the carbon emission of the type vehicles at the i-th road segment node at the T-th time instant, EF t denotes the carbon emission of the type vehicles at the i-th road segment node at the T-th time instant, EF type,speed,AirTem,Rh denotes the carbon emission factor of the type vehicles at the running speed, air temperature, and relative humidity. based on the energy consumption standard dataset and the carbon emission standard dataset, an output sample set is obtained and 4. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the dynamic estimation method of the energy consumption and carbon emission of the electric hybrid vehicle flow based on the space-time graph network in any one of claims 1 to 3.

5. A computer-readable storage medium storing a computer program, the computer-readable storage medium being characterized by, The computer program is executed by the processor to execute the dynamic estimation method of the energy consumption and carbon emission of the electric hybrid vehicle flow based on the space-time graph network in any one of claims 1 to 3.

Images