A method and device for evaluating the impact of built environment on the operation status of transportation system

Abstract

The present application discloses an evaluation method and device for the impact of the built environment on the operating state of a traffic system, which is used to solve the following technical problem: the existing evaluation method cannot accurately characterize the impact of the built environment on the operating state of the traffic system. The method includes: constructing a spatial proximity relationship deep neural network SPDNN model, and obtaining the fused spatial distance corresponding to the built environment through the SPDNN model; constructing a spatial weighted atrous convolutional neural network SWACNN model; inputting the fused spatial distance into the SWACNN model to obtain the spatial weight corresponding to the built environment.

Description

Technical Field

[0001] The present application relates to the technical field of transportation systems, and in particular to a method and device for evaluating the impact of a built environment on the operating status of a transportation system. Background Art

[0002] There is a complex interactive relationship between the urban built environment and the transportation system. The coordinated development of the two systems is not only conducive to achieving sustainable urban development, but also helps to improve the operation level of the transportation system and alleviate traffic congestion. The so-called built environment refers to the artificial environment formed after the natural environment is transformed, and the operation status of the transportation system generally refers to the road traffic service level, vehicle speed, road congestion index, etc. Studies have shown that the built environment will have a certain impact on the operation status of the transportation system, and the degree of impact varies with spatial changes, that is, the impact of the built environment on the operation status of the transportation system has spatial heterogeneity.

[0003] Existing studies have mostly used the traditional Geographical Weighted Regression (GWR) model to capture spatial heterogeneity, but this model has three main limitations: first, the measurement of spatial distance is relatively simple, mostly using direct coordinate distance or geographic topological space distance, resulting in an inaccurate description of the complex distance relationship between entities; second, the kernel function mostly uses the Gaussian kernel function, which has a weak ability to depict the attenuation and fluctuation of spatial relationships, resulting in the irregular attenuation of the impact of the built environment on the operating status of the transportation system being concealed; third, the bandwidth of different variables is unified, resulting in the inability to capture the changes in the relationship between different independent variables and dependent variables at a spatial scale.

[0004] Although there are a series of improved models in other fields, such as the semi-parametric geographically weighted regression model and the multi-scale geographically weighted regression model, the limitations of the GWR model have not been completely improved. For example, although the semi-parametric geographically weighted regression model takes into account the influence of different independent variables on the dependent variable, it only divides the variables into two types: global variables and local variables, and still cannot fully capture the bandwidth characteristics of each independent variable; and although the multi-scale geographically weighted regression model takes into account the bandwidth characteristics of each independent variable, the initial bandwidth determination and bandwidth optimization are based on the GWR model, and its kernel function still describes the changes in spatial relationships too regularly.

[0005] Therefore, exploring and studying a more adaptive GWR improved model is conducive to more accurately depicting the impact of the built environment on the operating status of the transportation system. Summary of the invention

[0006] The embodiments of the present application provide a method and device for evaluating the impact of the built environment on the operating status of a transportation system, so as to solve the following technical problem: the existing evaluation methods cannot accurately characterize the impact of the built environment on the operating status of a transportation system.

[0007] On the one hand, an embodiment of the present application provides a method for evaluating the impact of the built environment on the operating status of a transportation system, the method comprising: constructing a spatial proximity deep neural network SPDNN model, and obtaining a fused spatial distance corresponding to the built environment through the SPDNN model; constructing a spatial weighted atrous convolutional neural network SWACNN model; and inputting the fused spatial distance into the SWACNN model to obtain a spatial weight corresponding to the built environment.

[0008] In one or more embodiments of the present application specification, the SPDNN model includes an input layer, a hidden layer and an output layer; wherein the input layer is used to input the rectangular coordinate distance and the topological space distance corresponding to the built environment; the hidden layer is used to fit the nonlinear relationship between the rectangular coordinate distance and the topological space distance; the output layer is used to output the fused space distance between the rectangular coordinate distance and the topological space distance.

[0009] In one or more embodiments of the present application specification, the hidden layer of the SPDNN model includes a hidden layer-a and a hidden layer-b; wherein the hidden layer-a is connected to the input layer and includes 6 neurons; the hidden layer-b is connected to the output layer and includes 3 neurons.

[0010] In one or more embodiments of the present application specification, the SWACNN model includes an input layer, a dilated convolution layer, a fully connected layer and an output layer; wherein the input layer is used to input the fused spatial distance; the dilated convolution layer is used to perform a convolution operation on the fused spatial distance to obtain a feature matrix; the fully connected layer is used to perform nonlinear mapping on the feature matrix; and the output layer is used to output the spatial weight corresponding to the built environment.

[0011] In one or more embodiments of the present application specification, the dilated convolution layer includes a dilated convolution layer-a, a dilated convolution layer-b and a dilated convolution layer-c; wherein the dilated convolution layer-a is connected to the input layer; the convolution kernel of the dilated convolution layer-a is 3×3, the dilation rate is 1, and the number of channels is 32; the convolution kernel of the dilated convolution layer-b is 3×3, the dilation rate is 2, and the number of channels is 16; the convolution kernel of the dilated convolution layer-c is 3×3, the dilation rate is 3, and the number of channels is 8.

[0012] In one or more embodiments of the present application specification, the fully connected layer includes a fully connected layer-a and a fully connected layer-b; wherein the fully connected layer-a is connected to the hole convolution layer-c, which contains 256 neurons; the fully connected layer-b is connected to the output layer; which contains 128 neurons; and the output layer contains 6 neurons.

[0013] In one or more embodiments of the present application specification, after constructing the spatial weighted hole convolutional neural network SWACNN model, the method further includes: integrating the SPDNN model and the SWACNN model into a geographic hole convolutional neural network weighted regression GACNNWR model; optimizing and training the GACNNWR model, specifically including: obtaining built system indicator data and traffic system operation status indicator data within a preset geographic range to form training data; dividing the training data into a training set, a validation set, and a test set; splitting the training set to obtain a number of sub-training sets, and optimizing the GACNNWR model through the several sub-training sets. The GACNNWR model is trained to complete a model iteration and obtain the output fitting traffic system operation status indicator data; the loss function value of the GACNNWR model obtained during the model iteration process is used to construct a loss function trend graph, and whether the GACNNWR model has converged is determined based on the loss function trend graph; if so, the test set and the validation set are input into the GACNNWR model to verify the generalization ability of the GACNNWR model; if the difference between the prediction results of the GACNNWR model for the test set and the validation set is less than the preset difference threshold, it is determined that the GACNNWR model training is completed.

[0014] In one or more embodiments of the present application specification, the method also includes: determining built environment indicators and traffic system operating status indicators; constructing a traffic system operating status evaluation model based on the traffic system operating status indicators; wherein the traffic system operating status evaluation model is implemented by a GACNNWR model; acquiring built environment indicator data, and predicting the built environment indicator data through the traffic system operating status evaluation model to obtain spatial weights corresponding to the built environment indicators; fitting the traffic system operating status evaluation indicator data through the spatial weights based on the built environment indicator data and preset regression coefficients; analyzing the traffic system operating status evaluation indicator data through preset analysis indicators to obtain the influence of the built system indicators on the traffic system operating status indicators.

[0015] In one or more embodiments of the present application specification, the built environment indicators include at least: population density, land use mix, intersection density, business district accessibility and bus stop density; the transportation system operation status indicators include at least: traffic efficiency, traffic safety and driving comfort.

[0016] On the other hand, an embodiment of the present application also provides an evaluation device for the impact of the built environment on the operating status of a transportation system, the device comprising: a processor; and a memory having executable code stored thereon, wherein when the executable code is executed, the processor executes a method for evaluating the impact of the built environment on the operating status of a transportation system as described above.

[0017] The embodiment of the present application provides an evaluation method and device for the impact of the built environment on the operating state of the traffic system, which has the following beneficial effects: using rectangular spatial distance measurement and geographic topological spatial distance measurement as model input, using multiple hidden layers to fit the nonlinear effects between different distances, constructing a spatial proximity relationship deep neural network SPDNN model, obtaining a more accurate distance expression, and overcoming the problem of single spatial distance description. Then, the spatial weight corresponding to the built environment is output through the constructed spatial weighted hole convolutional neural network SWACNN model, breaking through the limitation that the existing kernel function is limited to the Gaussian function and is not selectable, and the spatial weight calculation no longer depends on the bandwidth, thereby breaking through the limitation that the bandwidth of different independent variables of the existing GWR model is unified, and the GACNNWR model is constructed and trained through the SPDNN model and the SWACNN model, and then the trained GACNNWR model is used to predict and analyze the built space indicator data, clarifying the impact of the built space indicator on the operating state of the traffic system, thereby realizing the fusion expression of complex spatial distance and the detection of spatial non-stationarity. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0019] Figure 1 A flow chart of a method for evaluating the impact of a built environment on the operating status of a transportation system provided in an embodiment of the present application;

[0020] Figure 2 A GACNNWR model structure diagram for an application scenario provided in an embodiment of the present application;

[0021] Figure 3 A diagram of the GACNNWR model optimization training process in an application scenario provided in an embodiment of the present application;

[0022] FIG. 4 is a graph showing the regression coefficient R in an application scenario provided by an embodiment of the present application. 2 Statistical graphs;

[0023] Figure 5 A structural diagram of a device for evaluating the impact of a built environment on the operating status of a transportation system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0025] The technical solution proposed in the embodiments of the present application is described in detail below with reference to the accompanying drawings.

[0026] Figure 1 Provided in the embodiments of this application. Figure 1 As shown,

[0027] Step 101: construct a spatial proximity relationship deep neural network SPDNN model, and use the model to obtain the fusion spatial distance corresponding to the built environment.

[0028] The method for evaluating the impact of the built environment on the operating state of the transportation system proposed in the embodiment of the present application is essentially to analyze the spatial heterogeneity of the impact of the built environment on the operating state of the transportation system. The analysis process is achieved by constructing a suitable neural network model.

[0029] First, the embodiment of the present application will first construct a spatial proximity deep neural network SPDNN model, which is used to describe the fused spatial distance corresponding to the built environment. Its input is the spatial distance corresponding to the built environment, and its output is the fused spatial distance corresponding to the input spatial distance, thereby solving the problem of single distance description in existing models.

[0030] Specifically, the structure of the SPDNN model can be found in Figure 2 ,like Figure 2 As shown, the SPDNN model includes an input layer for inputting the spatial distance corresponding to the built environment. In order to solve the problem that the existing model only describes the spatial distance of the built environment, the input spatial distance can be the rectangular coordinate distance and the topological space distance; Figure 2 It can be seen that the SPDNN model also includes two hidden layers for fitting the nonlinear relationship between the input rectangular coordinate distance and the topological space distance, and then the fused spatial distance is obtained through the output layer of the SPDNN model. In an example of the present application, the input of the SPDNN model is the Euclidean distance proximity relationship and the geographic space topological proximity relationship. These two relationships are fused through the proximity relationship deep neural network, and the output of the SPDNN model is a comprehensive spatial proximity relationship.

[0031] Furthermore, if Figure 2As shown, the SPDNN model has two hidden layers, the first hidden layer contains 6 neural clouds, and the second hidden layer contains 3 neurons.

[0032] Step 102: construct a spatially weighted atrous convolutional neural network (SWACNN) model.

[0033] After building the SPDNN model, a SWACNN model needs to be built to obtain the spatial weight corresponding to the built environment through the input fusion spatial distance. The spatial weight can be used to fit the relationship between the built environment indicators and the traffic system operation status indicators, thereby realizing impact analysis.

[0034] Therefore, if Figure 2 As shown, the structure of the SWACNN model includes an input layer, a dilated convolution layer, a fully connected layer, and an output layer, wherein the input layer is used to input the fused spatial distance output by the SPDNN model, and the dilated convolution layer includes three layers for performing a convolution operation on the fused spatial distance to obtain a feature matrix. In an example of the present application, the convolution kernel of the dilated convolution layer-a is 3×3, the dilation rate is 1, and the number of channels is 32. The convolution kernel of the dilated convolution layer-b is 3×3, the dilation rate is 2, and the number of channels is 16. The convolution kernel of the dilated convolution layer-c is 3×3, the dilation rate is 3, and the number of channels is 8. The fully connected layer includes 2 layers (hidden layers), which are used to perform nonlinear mapping on the feature matrix to obtain spatial weights, and in an example of the present application, the number of neurons in the fully connected layer-a is 256, and the number of neurons in the fully connected layer-b is 128. The number of neurons in the output layer is 6 (including 5 independent variables and 1 constant term), which corresponds to the spatial weights of the regression coefficients of each built environment indicator.

[0035] It should be noted that the atrous convolution layer-a is connected to the input layer, the atrous convolution layer-c is connected to the fully connected layer-a, and the fully connected layer-b is connected to the output layer.

[0036] Step 103: Input the fused spatial distance into the SWACNN model to obtain the spatial weight corresponding to the built environment.

[0037] After constructing the SWACNN model, the fused spatial distance output by the SPDNN model is input to obtain the spatial weight corresponding to the built environment.

[0038] In one or more possible implementations of the embodiments of the present application, after constructing the spatial weighted hole convolutional neural network SWACNN model, the method further includes: integrating the SPDNN model and the SWACNN model into a geographic hole convolutional neural network weighted regression GACNNWR model. Since the GACNNWR model is integrated from the aforementioned two models, the structural design of the model is the same as that of the aforementioned two models, and its network structure is set as shown in the following table:

[0039]

[0040] Furthermore, after building the GACNNWR model, before using it for prediction, the GACNNWR model needs to be optimized and trained. The training process is as follows: Figure 3 shown. Figure 3 The following is a diagram of the GACNNWR model optimization training process in an application scenario provided in an embodiment of the present application. Figure 3 As shown, the training process of the GACNNWR model includes at least:

[0041] 1. The geographic atrous convolutional neural network weighted regression GACNNWR model training process begins.

[0042] 2. Dataset division: First, divide the overall data set into data set A and data set B according to the 80 / 20 principle, and then divide data set A into training set and validation set according to the 80 / 20 principle; data set B is used as the test set.

[0043] In one or more possible implementations of the embodiments of the present application, the overall data set is obtained based on the built system indicator data and the traffic system operation status indicator data within a preset geographical range, wherein the built environment indicators include at least: population density, land use mix, intersection density, business district accessibility and bus station density; the traffic system operation status indicators include at least: traffic efficiency, traffic safety and driving comfort. In addition, in an example of the present application, the preset geographical range can be selected as a range of 900m×900m.

[0044] 3. Hyperparameter initialization: Initialize the hyperparameters of the neural network according to the pre-set hyperparameters, including initializing the learning rate, number of iterations, Dropout, etc.

[0045] In one example of the present application, the Dropout hyperparameter is used to enhance the generalization ability of the model.

[0046] 4. Split into Mini-batches: Split the training set into multiple Mini-batches based on the divided training set and the pre-set Batch-size.

[0047] 5. Training the network: A mini-batch sample is input into the network model, and the fitting value of y and the loss function value are calculated through the forward propagation algorithm. Then, the network model parameters are continuously updated through the backward propagation algorithm. The network is iterated and trained in sequence until all mini-batches are trained. Then, one epoch completes the network training, that is, one epoch completes the iteration.

[0048] 6. Fitting index calculation: After each Epoch iteration, the model fitting index is calculated, that is, the fitting traffic system operation status index data.

[0049] 7. Epoch number judgment: judge whether the number of epochs has reached the upper limit. If not, shuffle the training set and continue training. Otherwise, check the loss function trend chart to judge whether the model has converged. If the model has converged, stop iteration.

[0050] If the model does not converge, increase the Epoch upper limit and continue to iterate the model training.

[0051] 8. Verify the generalization ability of the model: After the model iteration is completed, use the iteratively trained model to predict the test set, check the prediction effect of the test set, and verify the generalization ability of the model. If the prediction effect of the test set is equivalent to that of the validation set, the trained model has a strong generalization ability and the model training is completed. Otherwise, adjust the model structure and retrain the model.

[0052] 9. The training process of the geographic atrous convolutional neural network weighted regression GACNNWR model is completed.

[0053] In an example of this application, in order to verify the application effect of the GACNNWR model and the difference with other models, the regression coefficient (R 2 ), mean absolute error (MAE), root mean square error (RMSE) and modified Akaike information criterion (AICc) were used to verify the performance of the GACNNWR model.

[0054] Furthermore, after the GACNNWR model is trained, it can be used to predict the impact of the built environment on the operating status of the transportation system, that is, the impact of spatial heterogeneity.

[0055] Specifically, based on the three indicators of the operation status of the traffic system, a traffic system operation status evaluation model is constructed through the GACNNWR model; in an example of the present application, the constructed traffic system operation status evaluation model includes the following three: GACNNWR-TE corresponding to traffic efficiency, GACNNWR-TS corresponding to traffic safety, and GACNNWR-TC corresponding to driving comfort; then the built environment indicator data is obtained, and the built environment indicator data is predicted through the traffic system operation status evaluation model to obtain the spatial weights corresponding to the built environment indicators; finally, based on the built environment indicator data and the preset regression coefficient, the traffic system operation status evaluation index data is fitted through the spatial weights.

[0056] Finally, by presetting the analysis indicator, in an example of this application, the regression coefficient R 2 The indicators are analyzed to analyze the traffic system operation status evaluation indicator data to obtain the impact of the built system indicators on the traffic system operation status indicators. 2 Statistical chart. Figure 4 shows the impact of various indicators of the built environment on various indicators of the operating status of the transportation system.

[0057] From the analysis of Figure 4(a), we can see that the impact of various indicators of the built environment on traffic efficiency is as follows:

[0058] ① Population density

[0059] A positive regression coefficient of population density means that the greater the population density, the greater the traffic index, the more congested the traffic, and the lower the traffic efficiency; a negative value means that the greater the population density, the smaller the traffic index, the smoother the traffic, and the higher the traffic efficiency.

[0060] ②HHI

[0061] A positive regression coefficient of HHI indicates that the smaller the HHI, the smaller the traffic system operation index, that is, the higher the level of urban land diversification, the smoother the traffic; a negative regression coefficient of HHI indicates that the more single the urban land type, the smoother the traffic.

[0062] ③Intersection density

[0063] A positive regression coefficient of intersection density indicates that the greater the intersection density, the greater the traffic index and the more congested the traffic. A negative regression coefficient indicates that the smaller the intersection density, the greater the traffic index and the more congested the traffic.

[0064] ④Accessibility of business district

[0065] When the regression coefficient of business district accessibility is positive, it means that the greater the business district accessibility, the greater the traffic index, and the more congested the traffic. Negative values ​​are the opposite.

[0066] ⑤ Public transportation proximity

[0067] When the bus proximity regression coefficient is positive, it means that the bus station density is smaller, the traffic index is smaller, and the traffic is smoother. A negative value means that the bus station density is greater, the traffic index is larger, and the traffic is more congested.

[0068] From the analysis of Figure 4(b), we can see that the impact of various indicators of the built environment on traffic safety is as follows:

[0069] ① Population density

[0070] A positive population density regression coefficient indicates that the greater the population density, the more traffic accidents there are. A negative value indicates the opposite.

[0071] ②HHI

[0072] A positive regression coefficient of HHI indicates that the greater the HHI, the more traffic accidents there are, that is, the more single the land use, the more traffic accidents there are. A negative value indicates that the greater the degree of mixed land use, the more traffic accidents there are.

[0073] ③Intersection density

[0074] The regression coefficients of intersection density are all positive, indicating that there are more traffic accidents in areas with high intersection density. The main reason is that the increase in intersections increases the risk of traffic conflicts.

[0075] ④Accessibility of business district

[0076] The regression coefficients of business district accessibility are all positive, indicating that the greater the business district accessibility, the greater the traffic accidents.

[0077] ⑤ Public transportation proximity

[0078] The regression coefficients of the proximity of buses in the study area are all positive, indicating that the area with a high density of bus stops also has a high number of traffic accidents. One possible reason is that there is a large demand for pedestrians to get on and off buses and cross roads near bus stops, which increases the time pedestrians are exposed to risk environments. In addition, the entry and exit of buses will cause certain disturbances to traffic flow, which will also increase the risk of traffic accidents.

[0079] From the analysis of Figure 4(c), we can see that the influence of various indicators of the built environment on driving comfort is as follows:

[0080] ① Population density

[0081] A positive population density regression coefficient indicates that the greater the population density, the lower the driving comfort; a negative value indicates the opposite.

[0082] ②HHI

[0083] The positive regression coefficient of HHI indicates that the smaller the HHI, the better the driving comfort, that is, the area with a high degree of mixed land use has a higher driving comfort. The negative value indicates that the driving comfort is better in the area with a single land use.

[0084] ③Intersection density

[0085] The regression coefficients of intersection density are mostly positive, indicating that areas with high intersection density have large acceleration interference and low driving comfort.

[0086] ④Accessibility of business district

[0087] When the regression coefficient of business district accessibility is positive, it means that the greater the business district accessibility, the greater the acceleration interference and the lower the driving comfort. A negative value means that the smaller the business district accessibility, the lower the driving comfort.

[0088] ⑤ Public transportation proximity

[0089] The regression coefficient of bus proximity is negative, indicating that areas with high bus station density have low acceleration interference and high driving comfort. The development of public transportation can promote the stability of traffic flow and improve driving comfort.

[0090] The above is a method embodiment of the present application. Based on the same inventive concept, the present application also provides an evaluation device for the impact of the built environment on the operating state of the traffic system, and its structure is as follows: Figure 5 shown.

[0091] Figure 5 This is a structural diagram of an evaluation device for the impact of the built environment on the operating status of a traffic system provided in an embodiment of the present application. Figure 5 As shown, the device includes: a processor; and a memory, on which executable codes are stored. When the executable codes are executed, the processor executes the above-mentioned method for evaluating the impact of the built environment on the operating status of the traffic system.

[0092] In one or more possible implementations of the embodiments of the present application, the processor is used to construct a spatial proximity deep neural network SPDNN model, and obtain the fused spatial distance corresponding to the built environment through the SPDNN model; construct a spatial weighted atrous convolutional neural network SWACNN model; and input the fused spatial distance into the SWACNN model to obtain the spatial weight corresponding to the built environment.

[0093] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0094] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity 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, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0095] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for evaluating the impact of built environment on the operation status of a transportation system, characterized in that: The method comprises: Constructing a spatial proximity relationship deep neural network SPDNN model, and obtaining the fusion space distance corresponding to the built environment through the SPDNN model; Construct a spatially weighted atrous convolutional neural network (SWACNN) model; Inputting the fused spatial distance into the SWACNN model to obtain the spatial weight corresponding to the built environment; The method further comprises: Determine indicators of built environment and transportation system operating status; Based on the traffic system operation status indicators, a traffic system operation status evaluation model is constructed; wherein the traffic system operation status evaluation model is implemented by a geographic atrous convolutional neural network weighted regression GACNNWR model, and the GACNNWR model is obtained by integrating the SPDNN model and the SWACNN model; Acquire built environment indicator data, and predict the built environment indicator data through the transportation system operation status evaluation model to obtain the spatial weight corresponding to the built environment indicator; Based on the built environment index data and the preset regression coefficient, the traffic system operation status evaluation index data is fitted through the spatial weight; By presetting analysis indicators, the traffic system operation status evaluation indicator data is analyzed to obtain the impact of the built system indicator on the traffic system operation status indicator.

2. The method for evaluating the impact of built environment on the operation status of a transportation system according to claim 1, characterized in that: The SPDNN model includes an input layer, a hidden layer and an output layer; wherein, The input layer is used to input the rectangular coordinate distance and the topological space distance corresponding to the built environment; The hidden layer is used to fit the nonlinear relationship between the rectangular coordinate distance and the topological space distance; The output layer is used to output the fused space distance between the rectangular coordinate distance and the topological space distance.

3. The method for evaluating the impact of built environment on the operation status of a transportation system according to claim 2, characterized in that: The hidden layers of the SPDNN model include hidden layer-a and hidden layer-b; wherein, The hidden layer-a is connected to the input layer and contains 6 neurons; The hidden layer-b is connected to the output layer and contains 3 neurons.

4. The method for evaluating the impact of built environment on the operation status of a transportation system according to claim 1, characterized in that: The SWACNN model includes an input layer, a hole convolution layer, a fully connected layer and an output layer; wherein, The input layer is used to input the fusion space distance; The atrous convolution layer is used to perform a convolution operation on the fused spatial distance to obtain a feature matrix; The fully connected layer is used to perform nonlinear mapping on the feature matrix; The output layer is used to output the spatial weight corresponding to the built environment.

5. The method for evaluating the impact of built environment on the operation status of a transportation system according to claim 4, characterized in that: The atrous convolution layer includes atrous convolution layer-a, atrous convolution layer-b and atrous convolution layer-c; wherein the atrous convolution layer-a is connected to the input layer; The convolution kernel of the dilated convolution layer-a is 3×3, the dilation rate is 1, and the number of channels is 32; The convolution kernel of the dilated convolution layer-b is 3×3, the dilation rate is 2, and the number of channels is 16; The convolution kernel of the dilated convolution layer-c is 3×3, the dilation rate is 3, and the number of channels is 8.

6. The method for evaluating the impact of built environment on the operation status of a transportation system according to claim 5, characterized in that: The fully connected layer includes a fully connected layer-a and a fully connected layer-b; wherein, The fully connected layer-a is connected to the atrous convolutional layer-c, which contains 256 neurons; The fully connected layer-b is connected to the output layer; it contains 128 neurons; and the output layer contains 6 neurons.

7. The method for evaluating the impact of built environment on the operation status of a transportation system according to claim 1, characterized in that: After constructing the spatial weighted atrous convolutional neural network (SWACNN) model, the method further includes: The GACNNWR model is optimized and trained, specifically including: Obtaining built system indicator data and traffic system operation status indicator data within a preset geographical range to form training data; Dividing the training data into a training set, a validation set, and a test set; The training set is divided into several sub-training sets, and the GACNNWR model is trained by the several sub-training sets to complete one model iteration, and the output fitting traffic system operation status indicator data is obtained; Constructing a loss function trend graph through the loss function value of the GACNNWR model obtained during the model iteration process, and judging whether the GACNNWR model has converged based on the loss function trend graph; If yes, the test set and the validation set are input into the GACNNWR model to verify the generalization ability of the GACNNWR model; If the difference between the prediction results of the GACNNWR model for the test set and the validation set is less than a preset difference threshold, it is determined that the training of the GACNNWR model is complete.

8. The method for evaluating the impact of built environment on the operation status of a transportation system according to claim 7, characterized in that: The built environment indicators include at least: population density, land use mix, intersection density, business district accessibility and bus stop density; the transportation system operation status indicators include at least: traffic efficiency, traffic safety and driving comfort.

9. An evaluation device for the impact of built environment on the operation status of a transportation system, characterized in that: The device comprises: processor; And, a memory having executable codes stored thereon, which, when executed, causes the processor to execute a method for evaluating the impact of a built environment on an operating state of a traffic system as described in any one of claims 1-8.

Images