Methodology for assessing the resilience of subway travel behavior during rainstorms, taking into account the built environment impact.

Abstract

This invention discloses a method for assessing the resilience of subway travel behavior during rainstorms, taking into account the impact of the built environment. It belongs to the field of urban transportation resilience assessment. The specific steps include: S1, collecting subway travel data, meteorological data, and built environment data around subway stations under rainstorm and ideal weather conditions; S2, quantifying the resilience value of subway travel behavior using the resilience triangle method; S3, constructing a spatially weighted geographic random forest model, using built environment data as features and resilience value as the objective, establishing a nonlinear mapping relationship between the two, and outputting the spatial distribution characteristics of resilience; S4, using the model to analyze the marginal contribution of each built environment factor to the resilience value, outputting the feature importance ranking, and identifying key built environment factors. This invention can quantify the disturbance and recovery capacity of subway travel behavior during rainstorms, reveal the mechanism by which the built environment affects travel behavior resilience, and provide a scientific basis for urban transportation resilience planning and rainstorm emergency management.

Description

Technical Field

[0001] This invention belongs to the field of urban transportation resilience assessment technology, and in particular relates to a method for assessing the resilience of subway travel behavior during rainstorms, taking into account the impact of the built environment. Background Technology

[0002] Current research has made some progress in assessing urban transportation resilience, mainly focusing on resilience assessments at the network topology and system function levels. These methods can reflect the overall service capacity of subway networks under extreme events, but they are insufficient in characterizing the resilience of individual or group passenger travel behavior. In fact, the disturbance and recovery of travel behavior directly reflects the urban transportation system's response to extreme weather and is an important dimension for measuring the resilience of the transportation system.

[0003] On the other hand, existing resilience studies often overlook the impact of built environment factors surrounding subway stations. Population density, land use type, and spatial distribution of commercial and public service facilities in the area where a subway station is located significantly influence passenger travel decisions and recovery processes during heavy rain. However, traditional methods, which often employ linear regression or overall statistical models, struggle to reveal the nonlinear relationship and spatial heterogeneity between the built environment and travel behavior resilience, leading to inaccurate identification of key influencing factors.

[0004] Therefore, there is an urgent need for a new approach that considers the impact of built environment factors around subway stations, utilizes spatial statistics and machine learning models to accurately quantify the resilience of subway travel behavior under rainstorms, and identifies key built environment influencing factors, so as to provide a scientific basis for urban transportation planning and emergency management. Summary of the Invention

[0005] The purpose of this invention is to provide a method for assessing the resilience of subway travel behavior during rainstorms, taking into account the impact of the built environment. This method is conducive to accurately quantifying the resilience of subway travel behavior during rainstorms and identifying key built environment impact factors, thus providing a scientific basis for urban transportation planning and emergency management.

[0006] To achieve the above objectives, this invention provides a method for assessing the resilience of subway travel behavior during heavy rainstorms, taking into account the impact of the built environment. The method includes the following steps:

[0007] S1. Collect subway travel data, meteorological data, and built environment data around subway stations under heavy rain and ideal weather conditions;

[0008] S2. Based on the data collected in S1, the resilience value of subway travel behavior is calculated using the resilience triangle method.

[0009] S3. A spatially weighted geographic random forest model is adopted, with the built environment data collected in S1 as the feature variable and the subway travel behavior resilience value calculated in S2 as the target variable. The model is trained to construct a global mapping relationship model between the built environment factors around the subway station and the subway travel behavior resilience, and output the spatial distribution characteristics of the subway travel behavior resilience.

[0010] S4, Utilization The model analyzes the spatially weighted geographic random forest model trained in S3, calculates the marginal contribution of each built environment factor to the resilience value of subway travel behavior, outputs the feature importance ranking, and identifies the key built environment factors around subway stations.

[0011] Preferably, S1 includes: subway travel data including the subway passenger entry station, exit station, entry time, exit time, origin and destination route numbers, and hourly passenger flow of each subway station under heavy rain and ideal weather conditions;

[0012] Meteorological data includes the start and end times of the rainstorm and the hourly rainfall.

[0013] Data on the built environment around subway stations includes: housing prices, distance from the city center, population density, building density, station floor area ratio, road network density, number of bus stops, distance from the nearest bus stop, mixed land use rate, density of catering facilities, density of commercial facilities, density of scientific research and education facilities, and density of medical facilities.

[0014] Preferably, the process of calculating the resilience value of subway travel behavior using the resilience triangle method in S2 is as follows:

[0015] S21. Determine the baseline subway travel volume under ideal weather conditions. Minimum subway ridership during heavy rain ;

[0016] S22. Determine the time t1 when subway travel volume begins to decline due to the rainstorm, the time t2 when subway travel volume reaches its lowest point due to the rainstorm, and the time t3 when subway travel volume recovers to a stable state. Plot the system performance curve P. (t) ;

[0017] S23, Based on system performance curve P (t) Baseline subway trips under ideal weather conditions The resilience value R of subway travel behavior is quantified by calculating the area enclosed by the two within the time interval t1-t3. The calculation formula is as follows:

[0018] ;

[0019] By approximating with the area of ​​a resilience triangle, a simplified calculation formula is obtained:

[0020] .

[0021] Preferably, the specific process of S3 is as follows:

[0022] S31. Determine the input variables for the model, using the calculated resilience value of subway travel behavior as the dependent variable. , This represents the subway travel behavior resilience value of the i-th out-of-bag sample, with the built environment factors around the collected subway stations as independent variables. , , ..., , ..., Where g=1, 2, p, where p is the total number of built environment factors. Let g be the value of the g-th built environment factor corresponding to the i-th site;

[0023] S32. Construct a local random forest model for each subway station as the basic building block of the spatially weighted geographic random forest model. The input of the local random forest model is the data of the neighboring subway stations of the station. The bandwidth of the neighboring subway stations is determined by the adaptive kernel function. The bandwidth is a predetermined number of nearest neighbors that is dynamically adjusted with the spatial density of the station.

[0024] S33. Construct the spatial weight matrix W, and the elements in the spatial weight matrix... The expression for calculating using the Gaussian distance decay function is as follows:

[0025] ;

[0026] in, Let be the spatial weight of the j-th station relative to the i-th station in the subway network. Let be the reachable distance between the i and j subway stations calculated using the Dijkstra shortest path algorithm, and h be the bandwidth determined by S32.

[0027] S34. Based on the spatial weight matrix W, the local random forest model in S2 is trained by weighted bootstrapping and random selection of a subset of variables at each decision tree split node. The outputs of all trained local random forest models are aggregated by weighted averaging of station spatial weights to form a global mapping relationship model. The nonlinear prediction value of subway travel behavior resilience for a single subway station i is calculated, as shown in the following expression:

[0028] ;

[0029] For the i-th station The local model parameters corresponding to the g-th built environment, This is the error term;

[0030] S35. Based on the global mapping relationship model, output the spatial distribution characteristics of subway travel behavior resilience.

[0031] Preferably, in S34, when training the local random forest model, the mean squared error (MSE) is calculated using out-of-bag samples to evaluate the model performance; the out-of-bag samples are the neighboring site data that were not selected during the bootstrap sampling process; the formula for calculating the mean squared error (MSE) is:

[0032] ;

[0033] Where k is the number of samples outside the bag. Let be the subway travel behavior resilience value of the i-th out-of-bag sample. This represents the resilience prediction of the out-of-bag sample by the spatially weighted geographic random forest model.

[0034] The importance of each built environment factor is calculated by altering the MSE of the substitution variable. For the g-th built environment factor... The importance of built environment factors VI )satisfy:

[0035] ;

[0036] Where Q is the total number of decision trees in the local random forest. Let be the mean squared error of the out-of-bag samples of the t-th decision tree on the original data. For the t-th decision tree, when replacing the g-th built environment factor... Mean square error of out-of-bag samples;

[0037] For the g-th built environment factor The importance of built environment factors is determined by their relative importance. Quantization, expressed as follows:

[0038] .

[0039] Preferably, in S35, spatial distribution characteristics are visualized through a partial dependency graph, and individual built environment factors are represented. Partial dependency functions satisfy:

[0040] ;

[0041] in, This represents a spatially weighted geographic random forest model, where G is the g-th built environment factor. The target value, To remove Other built environment factors, It is the i-th station The observed values, where n is the total sample size.

[0042] Preferably, S4 includes:

[0043] S41, Confirm The model input data is based on the global mapping relationship model constructed by S3, and includes the original data of the built environment factors around each subway station and the observed values ​​of subway travel behavior resilience.

[0044] S42, Utilization The model calculates the marginal contribution of the g-th built environment factor to the predicted travel behavior resilience value of the i-th subway station. The expression is as follows:

[0045] ;

[0046] Where S represents the absence of Let S be a subset of the built environment factors, where |S| is the number of elements in subset S. This is the prediction function for the global mapping relationship model. This indicates that the factor has a positive marginal contribution to the resilience of the i-th site. This indicates a negative marginal contribution;

[0047] S43. Rank the output features by importance. For each built environment factor g, calculate its importance across all sites. The absolute value and mean of the values ​​are used as the global importance score. The formula is:

[0048] ;

[0049] Score based on global importance All built environment factors are ranked from highest to lowest importance to generate a ranking result of feature importance;

[0050] S44. Identify key built environment factors, rank them according to the importance of features in S43, and filter the global importance scores. The top 6 built environment factors are key built environment factors that influence the resilience of travel behavior around subway stations.

[0051] Preferably, in S42, The model uses the average prediction value of the global mapping relationship model. Based on the baseline, the predicted value of subway travel behavior resilience at station i is... and The difference, through Values ​​are fairly allocated to each built environment factor, satisfying:

[0052] ;

[0053] in, To ensure that all factors The sum of values ​​and the resilience prediction value of the i-th site Compared with the model's average predicted value The deviations are completely equal, achieving precise quantification of marginal contribution.

[0054] Preferably, in S43, the site set is divided according to spatial regions to reflect the spatial differences in feature importance, and region-specific importance scores are calculated. The formula is:

[0055] ;

[0056] in, The number of sites in the corresponding region, based on The system generates a ranking of the importance of region-specific features and identifies key built environment factors in different regions. These regions are divided based on the distance of the stations from the city center, including the city center, suburbs, and outlying areas. The stations in each region constitute a set of stations.

[0057] Preferably, in step S44, after identifying key built environment factors, the method further includes visualizing the local influence characteristics of the key factors through a dependency graph, wherein the dependency graph is based on the values ​​of the key factors. The horizontal axis corresponds to value Using the vertical axis as the ordinate, the threshold range for factors to make positive / negative contributions to toughness is determined by fitting the curve. If the fitted curve satisfies:

[0058] ;

[0059] Then [a, b] is the threshold range for the key factor to make a positive marginal contribution to resilience.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] 1. This invention introduces built environment factors around subway stations into resilience assessment and constructs a global mapping relationship model between built environment and travel behavior resilience. This model can reveal the spatial differences of different stations under rainstorm weather, thus overcoming the limitations of traditional research in failing to adequately characterize the impact of the built environment.

[0062] 2. This invention employs a spatially weighted geographic random forest model, which can simultaneously capture the complex nonlinear relationship between built environment factors and travel behavior resilience, and effectively characterize the heterogeneity of their impact as it changes with geographic location, significantly improving the accuracy and reliability of the assessment model.

[0063] 3. The present invention further introduces The model can quantitatively output the marginal contribution of each built environment factor to resilience values ​​and their importance ranking, and visualize their influence mechanisms and threshold ranges through dependency graphs. This not only improves the interpretability of the model results, but also provides direct decision-making basis for urban transportation planning and extreme weather emergency management.

[0064] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0065] Figure 1 This is an overall flowchart of the method for assessing the resilience of subway travel behavior during rainstorms, taking into account the impact of the built environment, provided in this embodiment of the invention.

[0066] Figure 2 This is a schematic diagram of the subway travel behavior resilience triangle quantification provided in an embodiment of the present invention;

[0067] Figure 3 This is a flowchart illustrating the spatial distribution characteristics of subway travel behavior resilience provided in an embodiment of the present invention.

[0068] Figure 4 This is a flowchart of the process for identifying key built environment factors around subway stations, provided in an embodiment of the present invention. Detailed Implementation

[0069] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the embodiments of the present invention and are not intended to limit the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments in this application without creative effort are within the scope of protection of this application. Examples of embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout.

[0070] It should be noted that the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, such that a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such processes, methods, products, or devices.

[0071] Similar labels and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0072] In the description of this invention, it should be noted that the terms "upper," "lower," "inner," "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship in which the product of this invention is usually placed when in use. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention.

[0073] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," and "connect" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0074] This invention proposes a method for assessing the resilience of subway travel behavior during rainstorms, taking into account the impact of the built environment. Figure 1 The specific steps include:

[0075] S1. Collect subway travel data, meteorological data, and built environment data around subway stations under heavy rain and ideal weather conditions.

[0076] The subway travel data includes entry and exit stations, entry and exit times, route numbers of origin and destination, and hourly passenger flow at each subway station; meteorological data includes the start and end times of rainstorms and hourly rainfall; and the surrounding built environment data of subway stations includes, but is not limited to, housing prices, distance from the city center, population density, building density, station floor area ratio, road network density, number of bus stops, distance from the nearest bus stop, mixed land use rate, density of catering facilities, density of commercial facilities, density of scientific research and education facilities, and density of medical facilities.

[0077] S2. Based on the data collected in S1, the resilience value of subway travel behavior is calculated using the resilience triangle method, as shown in the diagram below. Figure 2 As shown. The specific process is as follows:

[0078] S21. Determine the baseline subway travel volume under ideal weather conditions. Minimum subway ridership during heavy rain .

[0079] S22. Determine the time t1 when subway travel volume begins to decline due to the rainstorm, the time t2 when subway travel volume reaches its lowest point due to the rainstorm, and the time t3 when subway travel volume recovers to a stable state. Plot the system performance curve P.(t)。

[0080] S23, Based on system performance curve P (t) Compared with the baseline travel volume The resilience value R of subway travel behavior is quantified by calculating the area enclosed by the two within the time interval t1-t3. The calculation formula is as follows:

[0081] ;

[0082] By approximating with the area of ​​a resilience triangle, a simplified calculation formula is obtained:

[0083] .

[0084] S3. Using a spatially weighted geographic random forest model, a nonlinear mapping relationship is established between built environment factors and subway travel behavior resilience, and the spatial distribution characteristics of resilience are output (e.g., Figure 3 (As shown). The specific process is as follows:

[0085] S31. Determine the model input variables, using the subway travel behavior resilience value calculated in S2 as the dependent variable. (The resilience value of subway travel behavior of the i-th out-of-bag sample), with the built environment factors around the subway station collected by S1 as independent variables. , , ..., , ..., Where g = 1, 2, p, where p is the total number of built environment factors. Let g be the value of the built environment factor corresponding to the i-th site.

[0086] S32. Construct a local random forest model for each subway station as the basic building block of the spatially weighted geographic random forest model. The input to the local random forest model is the data of neighboring subway stations for that station. The bandwidth of the neighboring subway stations is selected by an adaptive kernel function. The bandwidth is a predetermined number of nearest neighbors that is dynamically adjusted according to the spatial density of the station. The adjustment rule is: the bandwidth in the dense station area is set to 10-15 nearest neighbor stations, and the bandwidth in the sparse station area is set to 25-30 nearest neighbor stations to ensure that the model has sufficient training time.

[0087] S33. Construct the spatial weight matrix W, and the elements in the spatial weight matrix... Calculated using the Gaussian distance decay function, the following conditions are met:

[0088] ;

[0089] in, Let be the spatial weight of the j-th station relative to the i-th station in the subway network. Let be the reachable distance between the i-th and j-th subway stations calculated using the Dijkstra shortest path algorithm, and h be the bandwidth determined by S32.

[0090] S34. Based on the spatial weight matrix W, a local random forest model is trained by weighted bootstrapping and randomly selecting a subset of variables at each decision tree split node. The outputs of all trained local forest models are aggregated by weighted averaging of station spatial weights to form a global mapping relationship model. The nonlinear prediction value of subway travel behavior resilience for a single subway station i is calculated, as shown in the following expression:

[0091] ;

[0092] in, For the i-th station The local model parameters corresponding to the g-th built environment, This is the error term.

[0093] In training the local random forest model, out-of-bag samples are used to calculate the mean squared error (MSE) to evaluate model performance. Out-of-bag samples refer to neighboring site data that were not selected during the bootstrap sampling process. The formula for calculating the mean squared error (MSE) is:

[0094] ;

[0095] Where k is the number of samples outside the bag. Let be the subway travel behavior resilience value of the i-th out-of-bag sample. This represents the resilience prediction of the out-of-bag sample by the spatially weighted geographic random forest model.

[0096] The importance of each built environment factor is calculated by altering the MSE of the substitution variable. For the g-th built environment factor... The importance of built environment factors VI )satisfy:

[0097] ;

[0098] Where Q is the total number of decision trees in the local random forest. Let be the mean squared error of the out-of-bag samples of the t-th decision tree on the original data. For the t-th decision tree, when replacing the g-th built environment factor... The mean square error of the out-of-bag samples.

[0099] For the g-th built environment factor The importance of built environment factors is determined by their relative importance. Quantization, expressed as follows:

[0100] ;

[0101] in, Presented as a percentage, with the sum of the relative importance of all built environments being 100%.

[0102] S35: Based on the global mapping relationship model, output the spatial distribution characteristics of subway travel behavior resilience. These spatial distribution characteristics are visualized through a partial dependency graph, and individual built environment factors are also included. Partial dependency functions satisfy:

[0103] ;

[0104] in, This represents a spatially weighted geographic random forest model, where G is the g-th built environment factor. The target value, To remove Other built environment factors, It is the i-th station The observed values, where n is the total sample size.

[0105] when for At that time, the spatially weighted geographic random forest model predicts the resilience of subway travel behavior at the i-th station.

[0106] S4: Utilize The model analyzes the marginal contribution of each built environment factor to resilience, outputs a ranking of feature importance, and identifies key factors (such as...). Figure 4 (As shown). The specific process is as follows:

[0107] S41: Confirm The model input data is based on the global mapping relationship model constructed in S3, and includes the original data of built environment factors around each subway station collected in S1. and the resilience observations of subway travel behavior calculated by S2 .

[0108] S42: Utilize The model calculates the marginal contribution of the g-th built environment factor to the predicted travel behavior resilience value of the i-th subway station, i.e., the contribution of this factor. value ,satisfy:

[0109] ;

[0110] Where S represents the absence of Let S be a subset of the built environment factors, where |S| is the number of elements in subset S. This is the prediction function for the S3 global mapping relationship model. This indicates that the factor has a positive marginal contribution to the resilience of the i-th site. This indicates a negative marginal contribution. Among them, The model uses the average prediction value of the S3 global mapping model. Based on the baseline, the predicted value of subway travel behavior resilience at station i is... and The difference, through Values ​​are fairly allocated to each built environment factor, satisfying:

[0111] ;

[0112] in, This formula ensures that all factors are The sum of values ​​and the resilience prediction value of the i-th site Compared with the model's average predicted value The deviations are completely equal, achieving precise quantification of marginal contribution.

[0113] S43: Output feature importance ranking; for each built environment factor g, calculate its importance across all sites. The absolute value and mean of the values ​​are used as the global importance score. The formula is:

[0114] ;

[0115] Score based on global importance All built environment factors are ranked from highest to lowest importance to generate a ranking result based on feature importance. Specifically, the site set is divided by spatial region to reflect spatial differences in feature importance, and region-specific importance scores are calculated. The formula is:

[0116] ;

[0117] in, The number of sites in the corresponding region, based on The system generates a ranking of the importance of region-specific features and identifies key built environment factors in different regions. These regions are divided into three categories based on the distance of the stations from the city center: city center (distance ≤ 5km), suburban area (5km < distance ≤ 15km), and outer suburban area (distance > 15km). The stations in each region constitute a set of stations.

[0118] S44: Identify key built environment factors, rank them according to the importance of features in S43, and filter the global importance scores. The top six built environment factors are considered key built environment factors influencing the resilience of travel behavior around subway stations. After identifying these key built environment factors, further analysis is needed... Dependency graphs visualize the local influence characteristics of key factors, with the dependency graph based on the values ​​of the key factors. The horizontal axis corresponds to value Using the vertical axis as the ordinate, the threshold range for factors to make positive / negative contributions to toughness is determined by fitting the curve. If the fitted curve satisfies:

[0119] ;

[0120] Then [a, b] is the threshold range for the key factor to make a positive marginal contribution to resilience.

[0121] In summary, this invention provides a method for assessing the resilience of subway travel behavior during heavy rain considering the impact of the built environment. By collecting subway travel data and meteorological data under heavy rain and ideal weather conditions, as well as built environment data around subway stations, the method employs the resilience triangle approach to calculate subway travel behavior resilience. A spatially weighted geographic random forest model is used to construct a mapping relationship model between built environment factors around subway stations and subway travel behavior resilience, outputting the spatial distribution characteristics of subway travel behavior resilience. The model calculates the marginal contribution of built environment factors around subway stations to the resilience of subway travel behavior, outputs the importance ranking of features, and identifies key built environment factors around subway stations. This invention can accurately quantify the degree of disturbance and recovery ability of rainstorm weather on subway travel behavior, and reveal the mechanism by which built environment factors around subway stations affect the resilience of subway travel behavior.

[0122] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0123] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0124] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0125] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operations to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for assessing the resilience of subway travel behavior in rainstorm weather considering the impact of built environment, characterized in that, Comprise: S1, collect subway travel data, meteorological data under heavy rain and ideal weather, and built environment data around subway stations; S2, based on the data collected in S1, the resilience triangle method is used to calculate the resilience value of subway travel behavior, including: S21, determining the subway baseline travel volume under ideal weather and the minimum subway travel volume under heavy rain ; S22, determining the time t1 when the subway travel volume begins to decline under the influence of heavy rain, the time t2 when the minimum subway travel volume under the influence of heavy rain, and the time t3 when the subway travel volume returns to a stable state, and drawing a system performance curve P (t) ; S23, based on the system performance curve P (t) and the subway baseline travel volume under ideal weather , the resilience value of subway travel behavior is quantified by calculating the area enclosed by the two in the time interval t1-t3; S3, adopt spatially weighted geographically weighted random forest model to train the model, build a global mapping relationship model of built environment factors around subway stations and subway travel behavior resilience, and output the spatial distribution characteristics of subway travel behavior resilience, including: S31, determine the model input variables, taking the calculated subway travel behavior resilience value as the dependent variable , , represents the subway travel behavior resilience value of the ith out-of-bag sample, taking the collected built environment factors around the subway station as the independent variable , ,… ,… , p is the total number of built environment factors, is the value of the gth built environment factor corresponding to the ith station S32, build a local random forest model for each subway station as a basic component unit of the spatially weighted geographically weighted random forest model, the input of the local random forest model is the adjacent subway station data of the station, the bandwidth is selected by adaptive kernel function, and the bandwidth is the predetermined number of nearest neighbors dynamically adjusted according to the spatial density of the station; S33, constructing a spatial weight matrix W, elements in the spatial weight matrix W are calculated by a Gaussian distance decay function, and the expression is as follows: By a Gaussian distance decay function calculation, the expression is as follows: ； is the spatial weight of the i-th station to the j-th station in the subway station, is the road network reachable distance between the i-th and j-th subway stations calculated by the Dijkstra shortest path algorithm, and h is the bandwidth determined by S32. S34, based on the spatial weight matrix W, the local random forest model in S2 is trained by weighted bootstrap sampling and random selection of variable subset at each decision tree splitting node, and the output of all trained local random forest models is aggregated by station spatial weight weighted average to form a global mapping relationship model, and the nonlinear prediction value of subway travel behavior resilience of single subway station i is calculated; S35, according to the global mapping relationship model, the spatial distribution characteristics of subway travel behavior resilience are outputted; S4、Utilization The model analyzes the spatially weighted geographically random forest model trained in S3, calculates the marginal contribution of each built environment factor to the subway travel behavior resilience value, outputs the feature importance ranking, and identifies the key built environment factors around the subway station. 2.The method of claim 1, wherein, S1 comprises: The subway travel data includes the entry station, exit station, entry time, exit time, route number of the starting point and terminal point of subway passengers under heavy rain and ideal weather, and the hourly passenger flow of each subway station; The meteorological data includes the start and end time of heavy rain weather and hourly rainfall; The built environment data around the subway station includes: house price, distance from city center, population density, building density, station volume rate, road network density, number of bus stations, distance from the nearest bus station, land mixed use rate, catering facility density, commercial facility density, scientific research and education facility density, and medical facility density. 3.The method of claim 2, wherein, The process of calculating the subway travel behavior resilience value in S2 includes: Quantify the subway travel behavior resilience value R, the calculation formula is: ； The area of the resilience triangle is used for approximation to obtain the simplified calculation formula: 。 4.The method of claim 3, wherein, The specific process of S3 includes: Calculate the nonlinear prediction value of subway travel behavior resilience of single subway station i, the expression is as follows: ； for the i-th station local model parameters corresponding to the g-th built environment factor, is an error term.

5. The method for assessing the resilience of subway travel behavior in rainstorm weather considering the built environment impact according to claim 4, characterized in that: In S34, when training the local random forest model, the out-of-bag sample is used to calculate the mean square error to evaluate the model performance; The out-of-bag sample is the adjacent station data not selected in the bootstrap sampling process; The calculation formula of mean square error MSE is: ； where k is the number of out-of-bag samples, is the metro travel behavior resilience value of the i-th out-of-bag sample, is the resilience prediction value of the out-of-bag sample for the spatially weighted geographically weighted forest model. The importance of each built environment factor is calculated by the change of MSE of the variable by substitution, for the gth built environment factor , the importance of the built environment factor VI( ) satisfies: ； where Q is the total number of decision trees in the local random forest, is the gth out-of-bag mean squared error of the gth decision tree on the original data, is the gth out-of-bag mean squared error of the gth decision tree after permuting the gth built environment factor. for the gth built environment factor The importance of the built environment factor is quantified by the relative importance , expressed as follows: 。 6. The method for assessing the resilience of subway travel behavior in rainstorm weather considering the built environment impact according to claim 5, characterized in that: In S35, the spatial distribution characteristics are visualized by partial dependence plots, and individual built environment factors whose partial dependence functions satisfy: ； in, This represents a spatially weighted geographic random forest model, where G is the g-th built environment factor. The target value, To remove Other built environment factors, It is the i-th station The observed values, where n is the total sample size.

7. The method for assessing subway travel behavior resilience in storm weather considering built environment impact according to claim 6, characterized in that S4 Comprise: S41, determine Model input data, based on the global mapping relationship model constructed in S3, input the original data of the built environment factors around each subway station and the observed values of subway travel behavior resilience. S42、Utilizing Model calculates the marginal contribution of the gth built environment factor to the predicted value of the i th subway station travel behavior resilience , the expression is as follows: ； wherein S is a subset of built environment factors not containing |S| is the number of elements in the subset S, is a prediction function of the global mapping relationship model, represents a positive marginal contribution of the factor to the i-th site resilience, represents a negative marginal contribution; S43. Output feature importance ranking: For each built environment factor, calculate its importance across all sites. The absolute value and mean of the values ​​are used as the global importance score. The formula is: ； by global importance score ranking all built environment factors from high to low by global importance score, generating a feature importance ranking result; S44, identifying key built environment factors, screening global importance scores according to the importance ranking of features in S43 The top 6 built environment factors are identified as the key built environment factors that affect the resilience of the surrounding travel behavior of the subway station.

8. The method for assessing the resilience of subway travel behavior in rainstorm weather considering the built environment impact according to claim 7, characterized in that: In S42, The average prediction value of the global mapping relationship model The difference between the resilience prediction value of the subway travel behavior of the i-th station and the benchmark And The difference between the resilience prediction value of the subway travel behavior of the i-th station and the benchmark The value is fairly distributed to each built environment factor, which satisfies: ； wherein, , ensuring that the sum of the values of the factors is exactly equal to the resilience prediction value of the i-th site and the model average prediction value deviation, achieving precise quantification of marginal contribution.

9. The method for assessing the resilience of subway travel behavior in rainstorm weather considering the built environment impact according to claim 8, characterized in that: In S43, the site set is divided according to a spatial region, a spatial difference of feature importance is embodied, and a region-specific importance score is calculated The formula is: ； wherein, is the number of sites in the corresponding region, based on generating a region-specific feature importance ranking that identifies key built environment factors for different regions, which are divided according to distance from the city center, including the downtown area, the near suburb area, and the far suburb area, and the sites in each region form a site set.

10. The method for assessing the resilience of subway travel behavior in rainstorm weather considering the built environment impact according to claim 9, characterized in that: In S44, after identifying key built environment factors, the method also includes visualizing the local influence characteristics of these key factors through a dependency graph. The dependency graph uses the values ​​of the key built environment factors on the horizontal axis and the corresponding values ​​on the vertical axis. The value is on the vertical axis. By fitting the curve, the threshold range in which factors contribute positively or negatively to toughness is determined. If the fitted curve satisfies: ； Then [a, b] is the threshold interval of the key built environment factor which produces positive marginal contribution to resilience.

Images