Urban road network emission dynamic evaluation method for large-scale road section scale

Abstract

The invention discloses an urban road network emission dynamic evaluation method for a large-scale road section scale, and belongs to the technical field of urban road network emission prediction. In order to accurately consider the comprehensive influence of multi-scale space factors on carbon emission, a road network topological structure is constructed, obtained emission historical data of all road sections are expanded, then a multi-scale space feature extraction module is constructed, and the multi-scale space feature extraction module comprises a district feature extraction layer, a road section feature extraction layer, a gating fusion layer, a time feature extraction module and an emission prediction module. And then urban road network emission dynamic prediction oriented to a large-scale road section scale is carried out based on the emission historical data of all road sections. According to the method, data of all levels are organically integrated, potential relations among the data are fully mined, spatial distribution differences of road carbon emission are comprehensively and meticulously reflected, a solid foundation is laid for carbon emission analysis and prediction, the characterization capability of spatial features of the road carbon emission is improved, and the method is suitable for popularization and application. And powerful support is provided for formulation of accurate prediction and effective emission reduction strategies.

Description

Technical Field

[0001] The present invention belongs to the technical field of urban road network emission prediction, and in particular relates to an urban road network emission dynamic evaluation method oriented to large-scale road sections. Background Art

[0002] In the context of global response to climate change, the importance and urgency of dynamic assessment of road carbon emissions have become increasingly prominent. Data released by the International Energy Agency (IEA) in 2023 showed that carbon emissions in the transportation sector accounted for about 24% of the total global energy-related carbon emissions, of which road transport accounted for more than 70%, and this proportion is still on the rise.

[0003] Road carbon emissions are not only affected by factors such as vehicle type and fuel quality, but are also closely related to road and surrounding environmental factors. For example, due to the frequent start and stop of vehicles on congested roads in cities, carbon emissions per unit time are 30% - 50% higher than those on smooth roads. In addition, different types of road surrounding environments, such as commercial areas, industrial areas, and residential areas, have a significant impact on traffic flow and driving patterns, thereby changing carbon emission levels.

[0004] At present, most assessments of road carbon emissions lack dynamism and comprehensiveness. According to a survey conducted by relevant research institutions in 2023, more than 60% of existing assessment methods only consider a single or a few factors, and cannot accurately reflect the comprehensive impact of multi-scale spatial factors and road network connectivity on carbon emissions. This leads to a lack of accurate data support when formulating energy-saving and emission reduction policies, making it difficult to effectively reduce road carbon emissions.

[0005] Existing road carbon emission research methods are obviously deficient in considering multi-scale spatial factors. On the one hand, there is a lack of in-depth and systematic integration of road attributes, such as road grade and peak congestion, and the impact of these factors on carbon emissions at different scales has not been fully explored. On the other hand, the impact of emission characteristics on emissions is a nonlinear effect that may have cross-cutting and interactive effects. In existing studies, they are often analyzed in isolation, making it difficult to reflect the comprehensive impact of their complex interactions on carbon emissions.

[0006] Conducting dynamic assessments of road carbon emissions and comprehensively and meticulously considering multi-scale spatial factors have become key tasks in responding to climate change and achieving sustainable development. Summary of the invention

[0007] The problem to be solved by the present invention is to accurately consider the comprehensive impact of multi-scale spatial factors on carbon emissions, and propose a dynamic assessment method for urban road network emissions at a large-scale road section scale.

[0008] To achieve the above object, the present invention is implemented through the following technical solutions: A method for dynamic assessment of urban road network emissions at a large-scale road section scale includes the following steps: S1. Construct a road network topology structure, define the traffic network based on the roads and the connection relationship between roads, collect the carbon emission characteristic data of the road section in the past t time steps and its corresponding road network topology data, and obtain the emission history data of all road sections; S2. Input the emission history data of all road sections obtained in step S1 into a 1×1 convolutional layer for data expansion to obtain emission history data with expanded features; S3. Construct a multi-scale spatial feature extraction module, including a district feature extraction layer, a road section feature extraction layer, and a gated fusion layer. Input the emission history data set after the expanded features obtained in step S2 into the district feature extraction layer and the road section feature extraction layer respectively to extract features to obtain district attention scores and road section attention scores. Then, input the obtained district attention scores and road section attention scores into the gated fusion layer for feature fusion to obtain emission feature data after spatial feature extraction. S4. Construct a temporal feature extraction module, design an attention score between time steps based on the Transformer architecture, perform temporal feature extraction on the emission feature data after spatial feature extraction obtained in step S3, and obtain the emission feature data after spatiotemporal extraction; S5. Construct an emission prediction module, construct a method for the impact of other road sections on the emission characteristic data obtained after spatiotemporal extraction in step S4, and obtain an urban road network emission prediction result for a large-scale road section.

[0009] Furthermore, the road network topology constructed in step S1 is recorded as G = (V', E', J'), where V' represents the set of regional road segments, covering all road segments in the study area; E' is used to represent the connection relationship between road segments; J' is the adjacency matrix J'∈R constructed from the Euclidean distance between roads. N*N , N is the number of elements in the set V of regional road sections, R represents the real number set, indicating that the elements in the adjacency matrix are all real numbers, which are used in mathematics to accurately describe the properties of the matrix and the value range of the elements. i The road section in the past t Carbon emission characteristic data of time steps x i,t ∈R,1≤ i ≤N, get the emission history data X of all road sections in ∈R t×N .

[0010] Furthermore, the expression of the expansion process of step S2 is: X = Conv (Xin ) ; Among them, Conv represents the convolution operation, X is the emission history data of all road sections after the expanded features, and X∈R C×T×N , C represents the number of feature channels; Then, the learnable spatial position embedding matrix is ​​used to learn the longitude and latitude of the road sections within the range. The emission history data of all road sections after position embedding is: X ', the expression of the embedding process is: X =F(X,Rarea); Where F(·) is a 1×1 convolutional layer, Rarea is the adjacency matrix represented by longitude and latitude coordinates, Rarea∈R N×N .

[0011] Furthermore, the specific implementation method of step S3 includes the following steps: S3.1. Constructing the region feature extraction layer: First, construct the region query subspace , region key subspace and the region value subspace , the expression is: ; ; ; in, , , They are the regional query subspaces , region key subspace and the region value subspace The learnable weight matrix of Calculate the i The regional attention score between road segments , the expression is: ; in, Matrix form representing past time; It is i The corresponding value of each road segment in the regional query subspace, d k is the preset parameter, A i is the first scaling parameter optimized through training, B i is the second scaling parameter optimized through training; Then for the i The regional attention scores between the road sections are weighted to obtain the adjustedi Regional attention score of road segment The expression is: ; in, It is Hadamard. is a learnable tensor, softmax Function that converts a real vector into a probability distribution; Introducing Re into the feature transformation layer LU The function performs preliminary nonlinear transformation, and the calculation expression is: ; in, is the feature transformation matrix; Then, the multi-layer perceptron generation function is used to process the regional features after the preliminary transformation to obtain the processed regional features The expression is: ; in, is the multi-layer perceptron generating function; Then for the i The regional attention scores of the road sections are fused with the processed regional features to obtain the improved i The regional global characteristics of the road section are expressed as: ; in, represents the feature concatenation operation, is the projection matrix; S3.2. Construct the road segment feature extraction layer, including constructing the i The road segment query subspace , segment key subspace and the segment value subspace , the expression is: ; ; ; in, , , They are road segment query subspaces , segment key subspace and the segment value subspace The learnable weight matrix of The road segment query subspace , segment key subspace Perform dot product operation to calculate the i The road segment attention score of each road segment , the expression is: ; S3.3. Construct gated fusion layer; First, dimension adaptive weight calculation, road section feature dimension adaptive weight The calculation formula is ; in, Function to get the dimensions of the input feature vector; Adaptive weight of area feature dimension The calculation formula is: ; The expression for setting the gate signal is: ; in, Indicates i The gate signal of each road section is The activation function is calculated and has a value range of [0,1], which is used to perform weighted selection processing on the input data; They are functions for linear transformation of road section features and area features, respectively, and map multi-dimensional feature vectors into one-dimensional vectors through specific weight matrices; Then, feature fusion is performed based on the gated signal to obtain the first i Emission characteristics data for each road section , the expression is: .

[0012] Furthermore, the specific implementation method of step S4 includes the following steps: S4.1. Designing attention scores between time steps based on the Transformer architecture , based on each x i,t The expression for data containing T time steps is: ; in, denoted as a learnable weight matrix for the temporal training query subspace, is represented as a learnable weight matrix in the time-key subspace, represents the corresponding weight vector; Then the attention scores between time steps and the value subspace between time steps are Fusion, add the layer normalization operation LN of the residual connection after feature extraction, and obtain the first i Emission characteristics data for each road section , the calculation expression is: .

[0013] Furthermore, the specific implementation method of step S5 includes the following steps: S5.1. Considering other road segment features for the spatiotemporal extraction i The emission characteristic data of each road section is affected, and the road section feature fusion layer is constructed. i The emission characteristic data of each road section is fused with its own characteristics and the characteristics of other road sections at the same time to obtain the fused feature vector The expression is: ; in, For other road features, represents the feature concatenation operation, For road section j To the road section i The Euclidean distance of j Relative to the i Other sections of the road; S5.2. Convert the two-dimensional input feature vector Map to high-dimensional space and construct a high-dimensional feature matrix , the expression is: ; in, is the kernel function; Then, the high-dimensional feature matrix is ​​added to the regularization term to obtain the urban road network emission prediction result for large-scale road sections, which is expressed as: ; in, express T +1 time i Emission forecast for urban road network of road sections, It is the whole from In matrix form, W is the sample weight matrix, I is the identity matrix.

[0014] Beneficial effects of the present invention: The method for dynamic evaluation of urban road network emissions at a large-scale road section scale described in the present invention breaks through the limitations of traditional research methods and is no longer limited to single-dimensional considerations. Instead, it comprehensively analyzes the rich information at the level of road basic attributes, surrounding environment, and traffic flow characteristics. In terms of road basic attributes, it covers factors such as roads themselves, which directly affect the energy consumption and emissions of vehicles during driving. Surrounding environmental factors are conducive to the fact that different regional functions will lead to significant differences in traffic demand and travel patterns, which in turn have different degrees of effect on carbon emissions. On this basis, the present invention uses scientific and reasonable data fusion and feature engineering techniques to carefully construct a multi-scale spatial feature data set. Through a rigorous data processing process, data at all levels are organically integrated, and the potential connections between data are fully explored. The spatial distribution differences of road carbon emissions and the complex influencing factors behind them are extremely comprehensively and meticulously reflected, laying a solid foundation for subsequent carbon emission analysis and prediction, which is one of the core technologies of the present invention.

[0015] The method for dynamic assessment of urban road network emissions at a large-scale road section scale described in the present invention uses a variety of advanced time series analysis methods to conduct in-depth trend and cycle analysis of carbon emission time series. In the trend analysis process, the overall change trend of carbon emissions over time can be accurately identified, whether it is rising, falling or fluctuating, it can be clearly presented. Cycle analysis helps us discover the hidden periodic laws in carbon emission data, such as whether there are seasonal fluctuations, differences between weekdays and weekends, etc. At the same time, this technology also has the ability to handle the impact of special events. Through effective processing technology, the effect of these special events on carbon emissions can be accurately quantified, so as to more comprehensively grasp the temporal change law of carbon emissions. This technology provides crucial time dimension information for the subsequent prediction model, which is a key link in improving prediction accuracy, making the prediction results more in line with the actual carbon emissions.

[0016] The method for dynamic emission assessment of urban road networks at a large-scale road section scale described in the present invention takes into account the complexity and multidimensionality of carbon emission data, and the selected model needs to have strong feature learning and pattern recognition capabilities. After the model is built, the model parameters are reasonably adjusted to continuously tap the potential of the model so that it can better adapt to the characteristics of carbon emission data. At the same time, cross-validation technology is used to divide and train the data set multiple times to effectively avoid overfitting of the model and improve the generalization ability of the model. A variety of evaluation indicators, such as mean square error, mean absolute error, determination coefficient, etc., are used to comprehensively evaluate the model performance from different angles to ensure the accuracy and reliability of the model.

[0017] The method for dynamic assessment of urban road network emissions at a large-scale road section scale described in the present invention ensures that the prediction model can fully utilize the extracted spatial and temporal features to achieve high-precision carbon emission prediction. This not only provides a powerful tool for the study of road carbon emissions, but also provides a scientific basis for relevant departments to formulate energy-saving and emission reduction policies and plan transportation infrastructure. Systematically integrating multi-source information to enhance the ability to characterize the spatial characteristics of road carbon emissions provides strong support for accurate prediction and the formulation of effective emission reduction strategies, which is of great significance to promoting the green transformation of the transportation industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 The present invention provides a flow chart of a method for dynamic emission assessment of urban road networks at a large-scale road section scale. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention more clear, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention, that is, the specific embodiments described are only part of the embodiments of the present invention, rather than all of the specific embodiments. The components of the specific embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations, and the present invention can also have other embodiments.

[0020] Therefore, the following detailed description of the specific embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents the selected specific embodiments of the present invention. Based on the specific embodiments of the present invention, all other specific embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0021] In order to further understand the content, features and effects of the present invention, the following specific implementation methods are given as examples, and the attached Figure 1 The detailed instructions are as follows:

[0022] Embodiment 1: A method for dynamic assessment of urban road network emissions at a large-scale road section scale includes the following steps: S1. Construct a road network topology structure, define the traffic network based on the roads and the connection relationship between roads, collect the carbon emission characteristic data of the road section in the past t time steps and its corresponding road network topology data, and obtain the emission history data of all road sections; Furthermore, the road network topology constructed in step S1 is recorded as G = (V', E', J'), where V' represents the set of regional road segments, covering all road segments in the study area; E' is used to represent the connection relationship between road segments; J' is the adjacency matrix J'∈R constructed from the Euclidean distance between roads. N*N , N is the number of elements in the set V of regional road segments, R represents a real number set, and the first i The road section in the past t Carbon emission characteristic data of time steps x i,t ∈R,1≤ i ≤N, get the emission history data X of all road sections in ∈R t×N .

[0023] Furthermore, V' covers all road sections in the study area. E' is used to represent the connection relationship between road sections. If two road sections are connected to each other, there is a corresponding connection relationship record in E'.

[0024] S2. Input the emission history data of all road sections obtained in step S1 into a 1×1 convolutional layer for data expansion to obtain emission history data with expanded features; Furthermore, the expression of the expansion process of step S2 is: X = Conv (X in ) Among them, Conv represents the convolution operation, X is the emission history data of all road sections after the expanded features, and X∈R C×T×N , C represents the number of feature channels; Then, the learnable spatial position embedding matrix is ​​used to learn the longitude and latitude of the road sections within the range. The emission history data of all road sections after position embedding is: X ', the expression of the embedding process is: X =F(X, Rarea) Where F(·) is a 1×1 convolutional layer, Rarea is the adjacency matrix represented by longitude and latitude coordinates, Rarea∈R N×N .

[0025] Furthermore, the convolution layer plays a vital role in the entire data processing system. Its core function is to cleverly and accurately integrate the dynamically changing position information into the input data. In the actual operation process, it will carefully calculate and process each part of the input data according to the pre-set convolution rules, and integrate the position information into the feature representation of the data in a way that can be effectively used by subsequent models through convolution operations with specific convolution kernels. Then the position-embedded data X' is input into the regional multi-head self-attention unit for spatial feature extraction.

[0026] S3. Construct a multi-scale spatial feature extraction module, including a district feature extraction layer, a road section feature extraction layer, and a gated fusion layer. Input the emission history data set after the expanded features obtained in step S2 into the district feature extraction layer and the road section feature extraction layer respectively to extract features to obtain district attention scores and road section attention scores. Then, input the obtained district attention scores and road section attention scores into the gated fusion layer for feature fusion to obtain emission feature data after spatial feature extraction. For the calculation of the area feature extraction layer of road section i, it means taking road i as the center, and taking all connected road sections within a certain range as an area. The certain range is taken as a circular range with a radius of 5 kilometers, and the road sections within the range are circled through E' in the data and the corresponding connection relationship records.

[0027] Furthermore, the specific implementation method of step S3 includes the following steps: S3.1. Constructing the region feature extraction layer: First, construct the region query subspace , region key subspace and the region value subspace , the expression is: ; ; ; in, , , They are the regional query subspaces , region key subspace and the region value subspace The learnable weight matrix of Calculate the i The regional attention score between road segments , the expression is: ; in, Matrix form representing past time; It isi The corresponding value of each road segment in the regional query subspace, d k is the preset parameter, A i is the first scaling parameter optimized through training, B i is the second scaling parameter optimized through training; In the field of traditional traffic flow prediction and evaluation, the original formula maps the attention score of the road section to the interval [0,1] through the softmax function, and then multiplies it with the linearly transformed features to calculate the global features of the road section. However, emissions are different from traffic volume or speed. It is a process of continuous and stable change. It may be affected by the characteristics of surrounding roads and produce nonlinear and significant increases. This is because frequent acceleration and deceleration produces higher emissions than constant speed driving. The characteristics of the adjacent road network have complex nonlinear correlations. For example, the speed of vehicles decreases linearly when running in congestion guidelines, but at certain critical points, vehicles frequently stop and start, resulting in geometric changes in emissions.

[0028] In the calculation of attention scores, an extremely sophisticated weighting mechanism is introduced. By multiplying the attention score after softmax processing by the result of nonlinear transformation of the learnable tensor, this innovation changes the previous simple processing method, fully considers the differential impact of complex factors on carbon emissions in different sections, breaks through the limitations of traditional attention calculation, greatly enhances the accuracy of judging the importance of sections, and achieves more accurate focus on key sections, avoiding the omission of key information due to unreasonable attention allocation. Adjusted Regional attention score of section i: Then for the i The regional attention scores between the road sections are weighted to obtain the adjusted i Regional attention score of road segment The expression is: ; in, It is Hadamard. is a learnable tensor, softmax Function that converts a real vector into a probability distribution; At the feature transformation level, it was difficult to mine complex carbon emission-related features with a single linear transformation in the past. This improvement introduces nonlinearity with the help of the ReLU function and adds feature diversity through the generating function. This change takes into account the high complexity and nonlinearity of carbon emission characteristics, breaks through the simple thinking of traditional linear transformation, enhances the model's ability to learn complex features, enables the model to understand carbon emission data from a richer perspective, and will achieve more comprehensive and in-depth feature extraction, avoiding evaluation bias caused by insufficient feature extraction.

[0029] Introducing Re into the feature transformation layer LU The function performs preliminary nonlinear transformation, and the calculation expression is: ; in, is the feature transformation matrix; Then, the multi-layer perceptron generation function is used to process the regional features after the preliminary transformation to obtain the processed regional features The expression is: ; in, It is a multi-layer perceptron generating function; it generates a feature vector with the same feature dimension as the initial transformed feature and adds it to the initial transformed feature, breaking through the simple thinking of traditional linear transformation and enhancing the model's ability to learn complex features.

[0030] Then, in the dimension of information fusion, traditional methods often only focus on single-dimensional features, ignoring the synergy between features of different dimensions. This improvement uses a feature splicing method to splice the features with their projection transformation results and then multiply them with the adjusted attention score. This fully takes into account the comprehensive impact of multi-dimensional features on carbon emissions, breaks through the drawbacks of isolated information processing, enhances the model's ability to integrate multi-dimensional information, and will achieve a more comprehensive feature representation, avoiding one-sided evaluation caused by insufficient information fusion.

[0031] Then for the i The regional attention scores of the road sections are fused with the processed regional features to obtain the improved i The regional global characteristics of the road section are expressed as: ; in, represents the feature concatenation operation, is the projection matrix; S3.2. Construct the road segment feature extraction layer, including constructing the i The road segment query subspace , segment key subspace and the segment value subspace , the expression is: ; ; ; in, , , They are road segment query subspaces , segment key subspace and the segment value subspace The learnable weight matrix of Secondly, the correlation scores between sensor nodes are calculated. Since the core goal of feature extraction at the road level is to search for roads with similar features in the global scope, and then carry out feature extraction. In order to effectively reduce the amount of calculation without sacrificing accuracy, the sparse self-attention mechanism is used to realize the extraction of spatial features.

[0032] The road segment query subspace , segment key subspace Perform dot product operation to calculate the i The road segment attention score of each road segment , the expression is: ; S3.3. Construct gated fusion layer; The gated fusion layer plays a vital role in the entire model architecture. Its main function is to organically integrate dynamic features at multiple spatial scales and static spatial features with the help of the gating mechanism.

[0033] In the existing prediction and evaluation model architecture, the original gated fusion layer does not fully consider the dimensional differences between the road segment feature extraction layer and the area feature extraction layer when fusing multi-spatial scale features. Road segment features focus more on local details, while area features cover more macro information. This dimensional inconsistency makes it difficult to accurately weigh the importance of different features during gated calculations, resulting in insufficient information fusion.

[0034] This improvement overcomes this problem by introducing a dimensional adaptive adjustment mechanism to improve the calculation method of the gate signal. This mechanism can dynamically adjust the feature weights according to the dimensional characteristics of the road section and area features, avoiding information imbalance caused by dimensional differences and enhancing the model's ability to integrate features at multiple spatial scales.

[0035] First, dimension adaptive weight calculation, road section feature dimension adaptive weight The calculation formula is ; in, Function to get the dimensions of the input feature vector; Adaptive weight of area feature dimension The calculation formula is: ; The expression for setting the gate signal is: ; in, Indicatesi The gate signal of each road section is The activation function is calculated and has a value range of [0,1], which is used to perform weighted selection processing on the input data; They are functions for linear transformation of road section features and area features, respectively, and map multi-dimensional feature vectors into one-dimensional vectors through specific weight matrices; Then, feature fusion is performed based on the gated signal to obtain the first i Emission characteristics data for each road section , the expression is: .

[0036] S4. Construct a temporal feature extraction module, design an attention score between time steps based on the Transformer architecture, perform temporal feature extraction on the emission feature data after spatial feature extraction obtained in step S3, and obtain the emission feature data after spatiotemporal extraction; The extracted spatial features are integrated with the temporal features. It is necessary to extract time-dependent features from a time series consisting of traffic data at multiple time points. Time dependency covers space-time dependency. In view of the fact that relative position information is easily lost when extracting using traditional methods, a traffic time feature extraction module is proposed. It can selectively extract features based on traffic characteristics and relative position information of the time series.

[0037] The input sequence of each road section has a clear time sequence. In terms of time dependence, the input sequence presents a clear and orderly time sequence. This property plays a vital role in the accurate prediction of traffic data. However, the original Transformer architecture has certain limitations in learning the position information in the time series, and it is difficult to effectively capture this key information. In order to properly solve this problem, in the process of spatiotemporal feature extraction, the unique properties of the traffic system itself are closely combined to inject relative position information into the sequence. In addition, in the subsequent calculation of the attention score, a parameter that is continuously optimized through training is deliberately introduced. This parameter is specifically used to characterize the relative position.

[0038] Each x i,t Containing data of T time steps, considering the significant feature of traffic data with high temporal fluidity, that is, the time information of future moments will not affect the data change trend of previous time steps, corresponding weight matrices are generated for T different relative position relationships. Each road i is added with the relative position weight between roads.

[0039] Furthermore, the specific implementation method of step S4 includes the following steps: S4.1. Designing attention scores between time steps based on the Transformer architecture , based on each x i,t The expression for data containing T time steps is: ; in, denoted as a learnable weight matrix for the temporal training query subspace, is represented as a learnable weight matrix in the time-key subspace, represents the corresponding weight vector; In the process of traffic time feature extraction, after a series of preliminary calculations and processing steps, it enters the key time series dependency mining stage. At this time, the attention scores between all time steps It has been calculated that these scores reflect the closeness of the correlation between data at different time steps. At the same time, the corresponding value subspace Next, the attention score is multiplied by the value subspace. This multiplication operation is not a simple numerical calculation, but has profound significance. In essence, it is achieved by right The information in the feature extraction is weighted, selected and integrated to mine and determine the dependency relationship between time series. After feature extraction, a layer normalization operation LN with residual connection is added; Then the attention scores between time steps and the value subspace between time steps are Fusion, add the layer normalization operation LN of the residual connection after feature extraction, and obtain the first i Emission characteristics data for each road section , the calculation expression is: .

[0040] This operation effectively extracts the dependency characteristics between data points in each dimension. These characteristics are crucial for understanding the change pattern of traffic data over time and predicting future traffic conditions, enabling the model to more accurately capture and utilize the inherent laws in the time series, thereby improving the accuracy and reliability of the prediction.

[0041] S5. Construct an emission prediction module, construct a method for the impact of other road sections on the emission characteristic data obtained after spatiotemporal extraction in step S4, and obtain emission prediction results for urban road networks at a large-scale road section scale.

[0042] First, the road segment feature fusion layer is constructed. The carbon emission value of road segment i at time t is fused with its own characteristics and the characteristics of other road segments at the same time. The current method does not fully consider the influence of the road segment's own characteristics and the characteristics of other road segments. Through this step, the situation of only considering the carbon emission value in isolation is changed, and multiple aspects of information can be integrated, overcoming the problem of incomplete information utilization, and providing a richer feature basis for subsequent predictions.

[0043] Furthermore, the specific implementation method of step S5 includes the following steps: S5.1. Considering other road segment features for the spatiotemporal extraction i The emission characteristic data of each road section is affected, and the road section feature fusion layer is constructed. i The emission characteristic data of each road section is fused with its own characteristics and the characteristics of other road sections at the same time to obtain the fused feature vector The expression is: ; in, For other road features, represents the feature concatenation operation, Road Section j To the road section i The Euclidean distance of j Relative to the i Other sections of the road; S5.2. Convert the two-dimensional input feature vector Map to high-dimensional space and construct a high-dimensional feature matrix , the expression is: ; in, is the kernel function; Then, the high-dimensional feature matrix is ​​added to the regularization term to obtain the urban road network emission prediction result for large-scale road sections, which is expressed as: ; in, express T +1 time i Emission forecast for urban road network of road sections, It is the whole from In matrix form, W is the sample weight matrix, I is the identity matrix.

[0044] Furthermore, W is used to adjust the importance of different samples in the calculation, taking into account the different impacts of different samples on the prediction results, and avoiding the problem that all samples have the same impact weight on the results. is in inverse matrix form. By adding the regularization term , prevent the model from overfitting and ensure that the model still has good generalization ability under complex calculations.

[0045] for Forecast value, we need to talk about the forecast value Substitute the input parameters into the above steps and repeat the calculation.

[0046] It should be noted that relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0047] Although the present application has been described above with reference to specific embodiments, various modifications may be made thereto and parts thereof may be replaced with equivalents without departing from the scope of the present application. In particular, as long as there is no structural conflict, the various features in the specific embodiments disclosed in the present application may be used in combination with each other in any manner, and the fact that these combinations are not exhaustively described in this specification is only for the sake of omitting space and saving resources. Therefore, the present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A dynamic evaluation method for urban road network emissions at a large-scale road section scale, characterized in that: The steps include: S1. Construct a road network topology structure, define the traffic network based on the roads and the connection relationship between roads, collect the carbon emission characteristic data of the road section in the past t time steps and its corresponding road network topology data, and obtain the emission history data of all road sections; S2. Input the emission history data of all road sections obtained in step S1 into a 1×1 convolutional layer for data expansion to obtain emission history data with expanded features; S3. Construct a multi-scale spatial feature extraction module, including a district feature extraction layer, a road section feature extraction layer, and a gated fusion layer. Input the emission history data set after the expanded features obtained in step S2 into the district feature extraction layer and the road section feature extraction layer respectively to extract features to obtain district attention scores and road section attention scores. Then, input the obtained district attention scores and road section attention scores into the gated fusion layer for feature fusion to obtain emission feature data after spatial feature extraction. S4. Construct a temporal feature extraction module, design an attention score between time steps based on the Transformer architecture, perform temporal feature extraction on the emission feature data after spatial feature extraction obtained in step S3, and obtain the emission feature data after spatiotemporal extraction; S5. Construct an emission prediction module, construct a method for the impact of other road sections on the emission characteristic data obtained after spatiotemporal extraction in step S4, and obtain an urban road network emission prediction result for a large-scale road section.

2. According to claim 1, a large-scale section-scale urban road network emission dynamic assessment method is characterized in that: The road network topology constructed in step S1 is recorded as G = (V', E', J'), where V' represents the set of regional road segments, covering all road segments in the study area; E' is used to represent the connection relationship between road segments; J' is the adjacency matrix constructed from the Euclidean distance between roads J'∈R N*N , N is the number of elements in the set V of regional road sections, R represents the real number set, indicating that the elements in the adjacency matrix are all real numbers, which are used in mathematics to accurately describe the properties of the matrix and the value range of the elements. i The road section in the past t Carbon emission characteristic data of time steps x i,t ∈R,1≤ i ≤N, get the emission history data X of all road sections in ∈R t×N .

3. The method for dynamic evaluation of urban road network emissions at a large-scale road section scale according to claim 2 is characterized in that: The expression of the expansion process of step S2 is: X =Conv (X in ) ; Among them, Conv represents the convolution operation, X is the emission history data of all road sections after the expanded features, and X∈R C×T×N , C represents the number of feature channels; Then, the learnable spatial position embedding matrix is ​​used to learn the longitude and latitude of the road sections within the range. The emission history data of all road sections after position embedding is: X ', the expression of the embedding process is: X '=F(X,Rarity); Among them, F(·) is a 1×1 convolutional layer, Rarea is the adjacency matrix represented by longitude and latitude coordinates, and Rarea∈R N×N .

4. The method for dynamic emission assessment of urban road network at large-scale road section scale according to claim 3 is characterized in that: The specific implementation method of step S3 includes the following steps: S3.

1. Constructing the region feature extraction layer: First, construct the region query subspace , region key subspace and the region value subspace , the expression is: ； ； ； in, , , They are the regional query subspaces , region key subspace and the region value subspace The learnable weight matrix of Calculate the i The regional attention score between road segments , the expression is: ； in, Matrix form representing past time; It is i The corresponding value of each road segment in the regional query subspace, d k is the preset parameter, A i is the first scaling parameter optimized through training, B i is the second scaling parameter optimized through training; Then for the i The regional attention scores between the road sections are weighted to obtain the adjusted i Regional attention score of road segment The expression is: ； in, It is Hadamard. is a learnable tensor, softmax is a function that converts a real vector into a probability distribution; Introducing Re into the feature transformation layer LU The function performs preliminary nonlinear transformation, and the calculation expression is: ； in, is the feature transformation matrix; Then, the multi-layer perceptron generation function is used to process the regional features after the preliminary transformation to obtain the processed regional features The expression is: ； in, is the multi-layer perceptron generating function; Then for the i The regional attention scores of the road sections are fused with the processed regional features to obtain the improved i The regional global characteristics of the road section are expressed as: ； in, represents the feature concatenation operation, is the projection matrix; S3.

2. Construct the road segment feature extraction layer, including constructing the i The road segment query subspace , segment key subspace and the segment value subspace , the expression is: ； ； ； in, , , They are road segment query subspaces , segment key subspace and the segment value subspace The learnable weight matrix of The road segment query subspace , segment key subspace Perform dot product operation to calculate the i The road segment attention score of each road segment , the expression is: ； S3.

3. Construct gated fusion layer; First, dimension adaptive weight calculation, road section feature dimension adaptive weight The calculation formula is ； in, Function to get the dimensions of the input feature vector; Adaptive weight of area feature dimension The calculation formula is: ； The expression for setting the gate signal is: ； in, Indicates i The gate signal of each road section is The activation function is calculated and has a value range of [0,1], which is used to perform weighted selection processing on the input data; They are functions for linear transformation of road section features and area features, respectively, and map multi-dimensional feature vectors into one-dimensional vectors through specific weight matrices; Then, feature fusion is performed based on the gated signal to obtain the first i Emission characteristics data for each road section , the expression is: 。 5. The method for dynamic evaluation of urban road network emissions at a large-scale road section scale according to claim 4 is characterized in that: The specific implementation method of step S4 includes the following steps: S4.

1. Designing attention scores between time steps based on the Transformer architecture , based on each x i,t The expression for data containing T time steps is: ； in, denoted as a learnable weight matrix for the temporal training query subspace, is represented as a learnable weight matrix in the time-key subspace, represents the corresponding weight vector; Then the attention scores between time steps and the value subspace between time steps are Fusion, add the layer normalization operation LN of the residual connection after feature extraction, and obtain the first i Emission characteristics data for each road section , the calculation expression is: 。 6. The method for dynamic evaluation of urban road network emissions at a large-scale road section scale according to claim 5 is characterized in that: The specific implementation method of step S5 includes the following steps: S5.

1. Considering other road segment features for the spatiotemporal extraction i The emission characteristic data of each road section is affected, and the road section feature fusion layer is constructed. i The emission characteristic data of each road section is fused with its own characteristics and the characteristics of other road sections at the same time to obtain the fused feature vector The expression is: ； in, For other road features, represents the feature concatenation operation, For road section j To the road section i The Euclidean distance of j Relative to the i Other sections of the road; S5.

2. Convert the two-dimensional input feature vector Map to high-dimensional space and construct a high-dimensional feature matrix , the expression is: ； in, is the kernel function; Then, the high-dimensional feature matrix is ​​added to the regularization term to obtain the urban road network emission prediction result for large-scale road sections, which is expressed as: ； in, express T +1 time i Emission forecast for urban road network of road sections, It is the whole from In matrix form, W is the sample weight matrix, I is the identity matrix.

Images