A multi-objective optimization method for site selection of urban rail transit and related device

Abstract

This invention discloses a multi-objective optimization method and related apparatus for urban rail transit station site selection, belonging to the interdisciplinary field of urban planning and transportation engineering. The method includes the following steps: selecting intersections of urban roads of level 3 and above as candidate stations based on preprocessed urban road network data, and encoding the coordinates of the candidate stations; quantifying the service passenger flow, facility accessibility, and land development intensity of each candidate station based on preprocessed passenger travel origin-destination data, POI data, building data, and land use data, and forming a feature matrix for each candidate station; constructing a multi-objective transportation station site selection model; selecting candidate stations that meet distance constraints based on the feature matrix and using a greedy search algorithm; and inputting the selected candidate stations into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station site selection model to obtain the optimal planning result. This invention solves the problem of existing technologies having a single objective in site selection and failing to consider distance constraints between stations.

Description

Technical Field

[0001] This invention relates to the interdisciplinary field of urban planning and transportation engineering, specifically to a multi-objective optimization method and related apparatus for the site selection of urban rail transit stations. Background Technology

[0002] With the acceleration of urbanization, public transport-oriented development (PTD) has gradually become a major planning approach for solving urban congestion and community function allocation problems. This approach advocates for high-density development through mixed-use of rail transit stations and surrounding land, promoting the formation of pedestrian-friendly communities and thus driving sustainable urban development. However, practice shows that the benefits of rail transit stations exhibit significant spatial differentiation: high-quality stations can achieve a positive interaction through precise matching of land use and transportation demand, while inefficient stations become mere transportation nodes, resulting in resource waste.

[0003] Studies have shown that the benefits of rail transit stations depend on their location and the surrounding built environment. Current traditional subway station site selection studies are mostly based on a single objective, ignoring multi-dimensional collaborative optimization such as service population, facility accessibility and land use efficiency. This makes it difficult to balance the conflicts between multiple objectives and does not consider the minimum distance constraints between stations, which can easily lead to overlapping service areas or waste of resources. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-objective optimization method and related apparatus for the site selection of urban rail transit stations, so as to solve the problem that the existing technology has a single objective in site selection and does not consider the distance constraints between stations.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] Firstly, a multi-objective optimization method for urban rail transit station site selection includes the following steps:

[0007] Collect and preprocess urban road network data, population travel origin and destination data, POI data, building data, and land use data;

[0008] Intersections of urban roads of level 3 and above are selected as candidate stations based on the preprocessed urban road network data, and the coordinates of the candidate stations are encoded.

[0009] Based on the preprocessed population travel origin and destination data, POI data, building data and land use data, the service traffic, facility accessibility and land development intensity of each candidate site are quantified, and a feature matrix of each candidate site is formed;

[0010] A multi-objective transportation station location selection model is constructed. Based on the feature matrix, candidate stations that meet the distance constraints are selected using a greedy search algorithm. The results are then input into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station location selection model, thereby obtaining the optimal planning result.

[0011] In some implementations, the step of quantifying the service traffic, facility accessibility, and land development intensity of each candidate site based on the preprocessed population travel origin-destination data, POI data, building data, and land use data specifically includes:

[0012] Based on the preprocessed origin and destination data of the population, the inbound and outbound passenger flow of the candidate stations are extracted, and the inbound and outbound passenger flow are added together to obtain the service passenger flow.

[0013] Based on the preprocessed POI data, facility accessibility values ​​are calculated using an improved gravity model.

[0014] The preprocessed land use data includes the total area of ​​the plot and the area ratio of land within the plot. The preprocessed building data includes the building footprint and building height. The land use mixed entropy value is calculated based on the area ratio of land within the plot. The plot ratio of the candidate site is calculated based on the total area of ​​the plot, the building footprint, and the building height. The land use mixed entropy value and the plot ratio are multiplied to obtain the land use index of the candidate plot.

[0015] In some implementations, the multi-objective transportation station location model includes a multi-objective optimization function and constraints;

[0016] The multi-objective optimization function includes:

[0017]

[0018] The constraints include:

[0019]

[0020] Among them, Pop i For the service traffic of candidate site i, Acc i Landsue is the facility accessibility value for candidate site i. i Let x be the land use index of candidate site i. i Let s be the number of selected candidate sites, I be the set of candidate sites, and d be the number of candidate sites. ij D is the distance between candidate site i and candidate site j. min The minimum distance.

[0021] In some implementations, the step of filtering candidate sites that meet distance constraints based on the feature matrix and using a greedy search algorithm, and then inputting them into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station location model to obtain the optimal planning result, specifically includes:

[0022] After screening candidate stations that meet the distance constraints using a greedy search algorithm, the results are input into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station location model, and the Pareto front solution set is output.

[0023] Repeat the above steps until the maximum number of iterations is reached. Then, merge multiple Pareto front solution sets and perform non-dominated sorting to obtain the non-dominated solution as the global Pareto optimal solution, i.e., the optimal planning result.

[0024] In some implementations, the step of using a greedy search algorithm to filter candidate stations that meet distance constraints, then inputting them into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station location model, and outputting the Pareto front solution set, specifically includes:

[0025] Randomly select the candidate sites as the initial points. Based on the initial points and the feature matrix, use a greedy search algorithm to filter all candidate sites that meet the distance constraints as the input point set. The distance constraint is that the distance between two candidate sites is greater than or equal to a preset minimum distance.

[0026] The input point set is input into NSGA-II to initialize the population, and individuals are encoded using binary encoding. Selected candidate sites within an individual are encoded as 1, and unselected candidate sites are encoded as 0. The fitness value of an individual is calculated according to the multi-objective optimization function.

[0027] The population is non-dominated based on fitness values ​​to classify individuals into different Pareto front ranks, and the crowding degree of each individual in the same Pareto front rank is calculated.

[0028] Two individuals are randomly selected from the current population using the binary tournament selection method. The individual with a lower Pareto front level and greater crowding is selected as the parent, and two-point crossover and bit-flip mutation are performed sequentially to obtain the offspring.

[0029] The offspring and the current population are merged to obtain a new population. The new population is then sorted in a non-dominated manner. Individuals in the new population are divided into different Pareto front levels. After calculating the crowding degree of each individual in the same Pareto front level, the steps after randomly selecting two individuals are repeated until the maximum number of generations is reached and the Pareto front solution set is output.

[0030] In some implementations, the step of collecting and preprocessing urban road network data, population travel origin-destination data, POI data, building data, and land use data specifically includes:

[0031] Based on the road classification of urban road network vector data, duplicate roads and roads below the urban third-level road classification in the urban road network data are removed;

[0032] Based on the origin and destination data of the population, calculate the number of people leaving and entering the service area of ​​each pre-divided candidate station;

[0033] The POI data is classified and weighted according to the entropy of the number of POI data in different categories.

[0034] The building footprint and floor height are calculated based on building data, and the total area of ​​the plot and the proportion of land area within the plot are calculated based on land use data.

[0035] Secondly, a multi-objective optimization system for urban rail transit station site selection includes;

[0036] The data acquisition and preprocessing module is used to collect urban road network data, population travel origin and destination data, POI data, building data, and land use data and perform preprocessing.

[0037] The candidate site filtering module is used to filter the intersections of urban Class III roads and above as candidate sites based on the preprocessed urban road network data, and to encode the coordinates of the candidate sites.

[0038] The feature extraction module is used to quantify the service traffic, facility accessibility, and land development intensity of each candidate site based on the preprocessed population travel origin and destination data, POI data, building data, and land use data, and to form a feature matrix for each candidate site.

[0039] The multi-objective optimization module is used to construct a multi-objective transportation station location model. Based on the feature matrix, candidate stations that meet the distance constraints are selected through a greedy search algorithm and then input into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station location model to obtain the optimal planning result.

[0040] Thirdly, an electronic device includes a memory, a processor, and a computer program stored in the memory and executable in the processor, wherein the processor executes the computer program to implement the steps of the multi-objective optimization method for urban rail transit station site selection.

[0041] Fourthly, a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the multi-objective optimization method for urban rail transit station site selection.

[0042] Fifthly, a computer program product comprising a computer program that, when executed by a processor, implements the steps of the multi-objective optimization method for urban rail transit station site selection.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] This invention quantifies the service passenger flow, facility accessibility, and land development intensity of each candidate station by using preprocessed passenger travel origin-destination data, POI data, building data, and land use data, forming a feature matrix for each candidate station. After constructing a multi-objective transportation station site selection model, candidate stations are selected and the solution is obtained based on the feature matrix. Compared with existing technologies, this invention achieves multi-dimensional collaborative decision-making by simultaneously optimizing three core objectives: service passenger flow, facility accessibility, and land development intensity. This avoids planning imbalances caused by a single objective and overcomes the limitations of single-objective optimization. Furthermore, by using a greedy search algorithm to select candidate stations that meet distance constraints, it can directly filter out station combinations that do not meet the spacing requirements, avoiding the overlapping service areas or blind spots caused by excessive station density in traditional methods, which can lead to facility redundancy or inefficient operation.

[0045] Furthermore, by calculating the number of people leaving and entering each candidate site's service area using OD data, and taking the sum of the inbound and outbound passenger flow of the candidate sites as the service passenger flow, this invention can effectively overcome the shortcomings of traditional static analysis, which relies on fixed data and ignores the spatiotemporal characteristics of dynamic passenger flow.

[0046] Furthermore, this invention combines a local greedy search algorithm with the global genetic optimization NSGA-II algorithm. After merging multiple Pareto front solution sets, non-dominated sorting is performed to obtain the non-dominated solution as the optimal planning result. By using greedy search to quickly filter candidate sites that meet the distance constraints, the original candidate site set is compressed, significantly reducing the search space of NSGA-II, improving the convergence speed, and enabling the handling of large-scale candidate point sets. Attached Figure Description

[0047] Figure 1 The overall technical roadmap of the multi-objective optimization method for urban rail transit station site selection provided in the embodiments;

[0048] Figure 2 This is a schematic diagram of the spatial distribution of candidate sites;

[0049] Figure 3The spatial autocorrelation of candidate sites under different capability characteristics is shown in (a) the spatial distribution characteristics of the autocorrelation of facility accessibility value, (b) the spatial distribution characteristics of the autocorrelation of service traffic volume, and (c) the spatial distribution characteristics of the autocorrelation of land use index.

[0050] Figure 4 A schematic diagram of the SG-NSGA-II process provided for an embodiment;

[0051] Figure 5 This is a schematic diagram illustrating the site selection process for an example.

[0052] Figure 6 The diagram illustrates the changes in parameters, where (a) shows the changes in the average accessibility of facilities under different population sizes, (b) shows the changes in the average service traffic under different population sizes, and (c) shows the changes in the average land use index under different population sizes.

[0053] Figure 7 The diagram illustrates the changes in the number and performance of subway stations in the embodiment, where (a) shows the change in the average facility accessibility as the number of stations increases, (b) shows the change in the average service passenger flow as the number of stations increases, and (c) shows the change in the average land use index as the number of stations increases.

[0054] Figure 8 The following are spatial distribution diagrams for different numbers of stations in the embodiments, where (a) is a spatial distribution diagram for 150 stations, (b) is a spatial distribution diagram for 200 stations, (c) is a spatial distribution diagram for 250 stations, (d) is a spatial distribution diagram for 300 stations, (e) is a spatial distribution diagram for 350 stations, and (f) is a spatial distribution diagram for 380 stations.

[0055] Figure 9 The following is a schematic diagram comparing the optimal planning results with the original planning under different distance constraints in the example, where (a) is a schematic diagram of the original site planning, (b) is a schematic diagram of the optimal planning under the 800m distance constraint, and (c) is a schematic diagram of the optimal planning under the 1000m distance constraint.

[0056] Figure 10 A flowchart of a multi-objective optimization method for urban rail transit station site selection provided by the present invention;

[0057] Figure 11 This is a schematic diagram of a multi-objective optimization system for urban rail transit station site selection, provided as an example. Detailed Implementation

[0058] To enable those skilled in the art to better understand the present invention, the technical solution of the present invention will be further described in detail below with reference to the accompanying drawings. The content described herein is for explanation rather than limitation of the present invention.

[0059] It should be noted that the terms "comprising" and "having" and any variations thereof in the specification and claims of this invention are intended to cover a non-exclusive inclusion. For example, a process, method, system, product, or device 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, systems, products, or devices.

[0060] like Figure 1 and Figure 10 As shown in the figure, this embodiment provides a multi-objective optimization method for urban rail transit station site selection, including the following steps:

[0061] S1 collects and preprocesses urban road network data, population travel origin and destination data, POI data, building data, and land use data.

[0062] Specifically, anonymized mobile signaling data provided by China Unicom was collected on Thursday, April 15, 2021 (a weekday), covering the main urban area of ​​Xi'an (Weiyang District, Lianhu District, Yanta District, Baqiao District, and Chang'an District). The spatial resolution of the data is divided into 250-meter grids. Telecom operators determine an individual's origin and destination (OD) by viewing the locations where users stay in long-term trajectories, and determine the purpose of travel based on the time period of stay, such as work-home (WH), home-other (HO), etc. By summarizing the user's traffic flow between different grids, the flow of people entering and exiting the subway station area can be obtained. Therefore, the original population travel origin and destination data includes date (year, month, day), origin grid number (six digits), destination grid number (six digits), travel category, and number of people, as shown in Table 1.

[0063] Table 1. Sample data of population travel origin and destination points in mobile phone signaling data.

[0064]

[0065]

[0066] Furthermore, this technique utilizes multiple geospatial datasets to quantify the built environment surrounding subway stations. First, Point of Interest (POI) data is obtained from the public map service Amap (www.amap.com) to identify facilities around subway stations, thereby measuring accessibility from the station to these facilities. Urban land use data for Xi'an is obtained from the EULUC-China dataset to calculate land use data for the subway station area, including land use combinations. Amap provides building data for calculating floor area ratios. Urban road network data is also sourced from Amap.

[0067] Preferably, the preprocessing of the above data includes: (1) deleting duplicate roads and side streets in the urban road network data; (2) aggregating the flow of people with different travel purposes according to the grid of the origin and destination data of the people's travel, and calculating the number of people leaving and entering the service area of ​​each candidate station according to the origin and destination data of the people's travel; (3) classifying the POI data, and assigning weights to different categories of POIs according to the number of different categories of POIs based on the entropy weight method. Specifically, the POI data is divided into eight categories, namely CPOI (commercial), RPOI (residential), LPOI (leisure), IPOI (industrial), EPOI (education), MPOI (medical), GPOI (green landscape) and BPOI (bus stop). The number of different types of POIs varies greatly, so the entropy weight method is used to assign weights to different types according to the number of POIs to reduce the impact of the difference in magnitude, as shown in Table 2; (4) calculating the floor height and bottom area of ​​the building data and jointly encoding them with the coordinates.

[0068] Table 2 Weights of POIs in Different Categories

[0069]

[0070] S2, based on the preprocessed urban road network data, filter the intersections of urban third-level roads and above as candidate stations, and encode the coordinates of the candidate stations;

[0071] Specifically, ArcGIS Pro software was used to extract data from the preprocessed urban road network, and intersections of main urban roads (level III and above) were selected as candidate sites for subway stations. Taking Xi'an's main urban area as an example, a total of 1667 candidate points were collected. Figure 2 As shown.

[0072] All data in S1 and S2, including the data collection grid, are coordinate-encoded using the UTM-49N projection.

[0073] S3. Based on the preprocessed population travel origin and destination data, POI data, building data and land use data, the service population flow (Pop), facility accessibility (Acc), and land use intensity (Landuse) of each candidate site are quantified, and a feature matrix of each candidate site is formed.

[0074] Specifically, the service target capability characteristics of the 1667 candidate sites obtained in S2 are calculated. The service passenger flow, facility accessibility, and land development intensity of each candidate site are quantified.

[0075] (1) The method for calculating passenger flow is to sum the inflow and outflow of all grids within an 800-meter radius of each candidate station to obtain the inbound and outbound passenger flow, and then summarize the inbound and outbound passenger flow. The calculation formula is as follows:

[0076] Pop i =Inflow i +Outflow i

[0077] Among them, Pop i For the service traffic of candidate site i, Inflow i For the inbound traffic of candidate site i, Inflow i Let be the outbound passenger flow of candidate station i, where the inbound passenger flow and outbound passenger flow are the total inflow and total outflow of all grids within an 800-meter radius of candidate station i, respectively.

[0078] (2) Facility accessibility is obtained by integrating POI data from Gaode Maps and calculating it using an improved gravity model. Specifically, a distance decay function is used to simulate the accessibility from candidate sites to each POI facility, and a weighted summation is performed to generate the facility accessibility value for that candidate site, calculated using the following formula:

[0079]

[0080] Among them Acc i To implement the reachability value, W k Let f(d) be the entropy weight of the k-th type of POI. iP ) is the distance decay function, d ip Let be the distance from candidate site i to POI.

[0081] (3) Land development intensity is quantified by calculating the land use index using land use mixed entropy value and plot ratio. The diversity of land use is represented by the land use mixed entropy value, and the plot ratio is mainly used to represent the intensity of development, calculated by the following formula:

[0082]

[0083] Landuse i =Entropy i ·FAR i

[0084] Among them, Entropy i P represents the land use mixed entropy value within 800 meters of candidate site i. l FAR represents the area proportion of land parcel l within the plot. i Indicates the floor area ratio within 800 meters of candidate site i, Cover c floor indicates the building's footprint. c Indicates the building's floor height, A i Ladnduse represents the total area of ​​the plot at candidate site i. i This represents the land use index of candidate site i.

[0085] The above feature calculation method was used to summarize three features of 1667 candidate sites: service traffic, facility accessibility, and land development intensity, as shown in Table 3.

[0086] Table 3. Service Target Capability Characteristics of Candidate Sites

[0087]

[0088] In addition, the Moran I index was used for statistically significant spatial clustering and outlier analysis of the variables. Spatial characteristics were classified into four types: HH, LL, HL, and LH. HH (high-high clustering) indicates that candidate sites with high values ​​are surrounded by neighboring candidate sites with high values. LL (low-low clustering) indicates that candidate sites with low values ​​are surrounded by neighboring candidate sites with low values. High-low clustering (HL) and low-high clustering (LH) were labeled as outliers, representing either a candidate site with high values ​​being surrounded by neighboring candidate sites with low values, or a candidate site with low values ​​being surrounded by neighboring candidate sites with high values. Figure 3 The results of spatial autocorrelation clustering are presented. Red and blue dots represent candidate sites with strong spatial HH and LL clusters, yellow and green dots represent candidate sites with outliers (HL and LH), and the remaining white dots represent candidate sites with insignificant autocorrelation. It can be observed that the autocorrelation of the three measurement indicators (service passenger flow, facility accessibility value, and land use index) exhibits similar spatial distribution characteristics. Candidate sites in both the city center and suburbs show significant spatial autocorrelation, with high values ​​concentrated in the city center and low values ​​distributed in the city periphery, indicating that Xi'an's travel demand, accessibility, and land use development exhibit a monocentric spatial structure. Most candidate sites with insignificant autocorrelation are located between the city center and the periphery.

[0089] S4. Construct a multi-objective transportation station location model. Based on the feature matrix, filter candidate stations that meet the distance constraints using a greedy search algorithm, and then input them into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station location model to obtain the optimal planning result.

[0090] Specifically, in S4.1, the above problem is modeled to construct a multi-objective transportation station location model. This problem can be simplified to selecting the optimal combination of stations from the 1667 candidate stations, considering factors such as passenger flow, facility accessibility, and land development intensity. Using 0-1 coding to represent whether a candidate station is selected, this multi-objective transportation station location model can be established as an integer linear programming model, where the multi-objective optimization function includes:

[0091]

[0092] Maximize the above three objective functions, which represent the sum of the three capability features of the selected candidate sites;

[0093] The constraints include:

[0094] ∑ i∈I x i =s(candidate site number constraint)

[0095]

[0096] (Minimum distance constraint, D) min =800m)

[0097] Among them, Pop i For the service traffic of candidate site i, Acc i Landsue is the facility accessibility value for candidate site i. i Let x be the land use index of candidate site i. i Let s be the number of selected candidate sites, I be the set of candidate sites, and d be the number of candidate sites. ij D is the distance between candidate site i and candidate site j. min The minimum distance.

[0098] The proposed multi-objective transportation station site selection model considers three objective functions, aiming to select the optimal combination of subway station sites. The objective function maxf1 is to maximize the service passenger flow of all selected candidate sites; maxf2 is to maximize the facility accessibility of all selected candidate sites; and maxf3 is to maximize the land use index of all selected candidate sites. Constraints limit the number of candidate sites ultimately selected. Therefore, to maximize the three objective functions of all selected candidate sites with a specific number of sites, the optimal combination of subway stations must be selected.

[0099] S4.2 uses a greedy strategy to pre-select candidate sites, satisfying the method's constraint on the distance between two candidate sites. This embodiment proposes a Spatiallimited Greedy Fast Non-dominated Sorting Genetic Algorithm II (SG-NSGA-II) based on a spatially constrained greedy strategy to solve the multi-objective subway station site selection problem. By introducing local greedy search techniques and adaptive control mechanisms, the algorithm's optimization efficiency and solution quality are improved. The algorithm's process includes: main iterative loop, pre-screening process based on greedy search, solution process based on NSGA-II, and final solution refinement. The entire process is described as follows: Figure 4 As shown, the solution process is as follows: Figure 5 As shown:

[0100] (1) Main Iteration Loop

[0101] The main loop iterates until the set maximum number of iterations n. In each iteration, the algorithm starts from a new random initial point and uses a greedy search algorithm to select candidate stations that meet the distance constraints as input points to NSGA-II. The algorithm solves the multi-objective optimization function of the multi-objective transportation station location model until the maximum number of iterations is reached, at which point a global Pareto optimal solution is generated. Each main loop will generate a set of Pareto optimal solutions.

[0102] (2) Greedy search process

[0103] like Figure 4 As shown, the process will execute s loops. Each main loop's greedy search starts from a new random initial point and then progressively searches for points that satisfy the distance constraint until no other points meeting the requirement can be found. The result of this process will serve as the input point set for NSGA-II. The distance constraint is that the distance between two candidate sites is greater than or equal to a preset minimum distance, which is D in S4. min =800m.

[0104] (3) NSGA-II solution process

[0105] The input point set is input into NSGA-II to initialize the population P, and individuals are encoded using binary encoding. Selected candidate sites within an individual are encoded as 1, and unselected candidate sites are encoded as 0. The fitness value of an individual is calculated according to the multi-objective optimization function.

[0106] The population is non-dominated based on fitness values ​​to classify individuals into different Pareto front ranks, and the crowding degree of each individual in the same Pareto front rank is calculated.

[0107] Genetic manipulation is performed using a binary tournament selection method, randomly selecting two individuals from the population. The quality of these two individuals is compared, and the individual with a lower Pareto front rank and higher crowding is selected as the parent. Two-point crossover and bit-flipping mutations are then performed sequentially to obtain offspring Q.

[0108] The offspring Q is merged with the current population P to obtain a new population R. A non-dominated sort is performed on the new population R, and individuals within R are divided into different Pareto front levels. The crowding degree of each individual at the same Pareto front level is calculated. The steps following the genetic operation are repeated, with the generation counter incremented by 1, and the loop continues until the maximum number of generations is reached, at which point the Pareto front solution set is output. If a Pareto front cannot be fully incorporated into the new generation (e.g., insufficient remaining slots), the individuals at that Pareto front are sorted from largest to smallest crowding degree, retaining those with the largest distance to ensure diversity.

[0109] (4) Determine the final solution

[0110] After the main loop ends, the Pareto front solution sets obtained in each loop are merged and sorted non-dominated to obtain a set of non-dominated solutions, which are used as the global Pareto optimal solution, i.e. the optimal planning result.

[0111] S5, NSGA-II uses a genetic algorithm to solve the multi-objective transportation station location problem mentioned above, requiring adjustment of its hyperparameters to obtain optimal performance. These three objective functions will serve as the basis for evaluating the model, thereby determining the parameter values. Since each generation produces a Pareto front solution set, the average value of the objective function of the Pareto front solution set is calculated to represent the performance of each generation. The maximum number of generations in the genetic selection is set to 1000, which, empirically, is sufficient to ensure convergence. The convergence point of the function determines the value of the generation number parameter. Furthermore, this embodiment uses a grid search method to explore the optimal population size. This embodiment sets a proportionally increasing population size {50, 100, 150, 200}, and selects the optimal population size by comparing the objective function values ​​under different population sizes. Figure 6The generational changes of the three objective functions under different parameters are shown. Analysis of the three objective functions reveals that population size significantly impacts the performance of the NSGA-II algorithm. Larger populations (150 or 200) tend to explore the search space better, thus improving convergence speed and the final objective values ​​of the three objective functions.

[0112] Specifically, the final objective function value with a population size of 50 is lower than that of the other parameters, indicating that this parameter ultimately failed to achieve optimal performance. While a population size of 100 eventually approaches the optimal objective function value, the convergence speed is too slow, requiring an additional 200+ generations of genetic evolution. Setting the population size to 150 is the best choice to balance computational efficiency and optimization performance. It exhibits fast convergence, high-quality solutions, and stability across all three objective functions. It is worth noting that while a population size of 200 provides the best final performance, the trade-off in computational cost is also important, especially for problems with high function evaluation costs. A population size of 150 achieves near-optimal performance with lower computational overhead, making it the preferred choice for practical applications. Ultimately, this embodiment selects 500 as the maximum number of generations for the function to converge to the optimal value, and a population size of 150.

[0113] S6, change the distance constraints of the candidate stations to solve the problem, and change the minimum distance constraint D. min The distance is set to 1000m, and step S4.2 is repeated to obtain the site selection situation (optimal planning result) under the 1000m distance constraint. The three objective function values ​​of the two site selection situations under the 800m and 1000m distance constraints and the original site selection situation are calculated and compared.

[0114] The SG-NSGA-II algorithm described above is used to continuously calculate the overall performance under different numbers of candidate sites. Then, the average of the objective function of a set of solutions on the Pareto front is calculated. The number of candidate sites is determined by observing how the average converges as the number of candidate sites changes. Specifically, when the performance improvement of candidate sites is not significant or the improvement rate slows down significantly, this method considers that the number of candidate sites is sufficient, because the performance of the remaining candidate sites is poor.

[0115] Figure 7 The graph illustrates how the three objective functions change with the number of candidate stations. The convergence of these functions is clearly visible. Starting from approximately 320 candidate stations, the change in the objective functions begins to slow. From 380 candidate stations, the objective functions gradually plateau, even showing a slight decrease in passenger flow and land use index. Therefore, it can be concluded that when the number of stations reaches 380, the main service capacity of Xi'an's subway stations converges, and the number of subway stations approaches saturation.

[0116] Figure 8 This study demonstrates the spatial distribution characteristics and trends of different station numbers. Specifically, the overall spatial distribution of subway stations is independent of the number of stations. The optimal spatial location of subway stations exhibits the same distribution trend as the city's population, facilities, and land use. This is determined by the objective function of this embodiment. Specifically, the location selection of subway stations gradually spreads from the city center to the city periphery. Meanwhile, the northwest corner of the city, being the site of the Han Dynasty imperial palace, Daming Palace, has not had a subway station planned due to government considerations for its protection. The southwest and further south areas of the city will be the first to have subway lines opened. These two areas are the High-tech Zone and the Chang'an District University Town, as employment opportunities and educational institutions are highly concentrated there. Maps with 300 and 350 stations show a vacant area in the southwest, where the main land use types are parkland and industrial land, with fewer pedestrians and facilities. Therefore, this embodiment can understand the planning location and priority order of subway stations from these analyses.

[0117] S7. To test the performance of the SG-NSGA-II method on the spatial location optimization problem of subway stations, this method compares the optimization results with the existing urban subway plan of Xi'an. To strengthen the comparison, the method also calculates the average distance of existing stations in Xi'an as 1080.99 meters. Based on the planning recommendations, two distance constraints of 800m and 1000m are set to implement the method. Currently, the total number of planned subway stations in the study area is 167, and the number of subway stations in the algorithm is set to 167. After 100 greedy search loops, the population size of NSGA-II is set to 150, and the number of generations in each loop is set to 500. Then, non-dominated screening is performed on 100 Pareto solutions to form the final Pareto front (…). Figure 9 To compare the specifics of the three schemes, the method randomly selected one of the Pareto optimal schemes and visualized it against the original site. Figure 9 The spatial distribution of the three schemes is shown, and Table 4 shows the specific performance values ​​of the three schemes.

[0118] Table 4 Comparison of optimization results in this embodiment with the original site

[0119]

[0120] Figure 9As can be seen, the spatial distribution of the three schemes shows roughly the same trend, which is consistent with the spatial agglomeration of population and facilities in Xi'an. This also indicates that the three schemes follow similar planning principles and are reasonable. Specifically, the scheme under the 800-meter distance constraint has a denser spatial distribution than the other two schemes. The scheme under the 1000-meter distance constraint has a density similar to existing stations, but a more uniform spatial distribution, indicating that the distance constraint between stations is a significant factor influencing station location. Table 4 shows the specific performance improvements of the SG-NSGA-II method provided in this embodiment for the optimal subway station location problem. The results show that under the 800-meter distance constraint, the SG-NSGA-II method significantly improves the efficiency of subway stations, increasing facility accessibility by 15.337%, service passenger flow by 16.378%, and land development intensity by 38.282%. Under the 1000-meter distance constraint, land development intensity also increases by 21.4158%, but facility accessibility and service passenger flow decrease slightly, mainly due to the uneven distribution of population and facilities. This also indicates that the current station density in Xi'an is insufficient, with some stations being too far apart, failing to maximize service efficiency. Xi'an urgently needs to increase the number of stations to better meet urban transportation demands. This also demonstrates the superiority and reliability of the multi-objective optimization method for urban rail transit station site selection provided in this embodiment in practical urban subway station site selection.

[0121] Therefore, this embodiment has the following advantages: (1) It can provide a variety of optimization schemes through Pareto front solution set, and support decision-makers to weigh the priority of objectives according to their needs; (2) Spatial constraints are precisely controlled, and greedy pre-screening ensures that the distance between candidate sites is ≥800m, avoiding resource waste and improving planning feasibility; (3) Algorithm efficiency: SG-NSGA-II combines local greedy search and global genetic optimization to obtain the location benefits of the final selected planning sites, which are improved by 15%-40% in three aspects; (4) Data-driven decision-making: It integrates multi-source data such as mobile phone signaling and POI to quantify the dynamic characteristics of pedestrian flow and facility distribution, and enhance the scientific nature of planning.

[0122] like Figure 11 As shown, this embodiment provides a multi-objective optimization system for urban rail transit station site selection, including:

[0123] The data acquisition and preprocessing module is used to collect urban road network data, population travel origin and destination data, POI data, building data, and land use data and perform preprocessing.

[0124] The candidate site filtering module is used to filter the intersections of urban Class III roads and above as candidate sites based on the preprocessed urban road network data, and to encode the coordinates of the candidate sites.

[0125] The feature extraction module is used to quantify the service traffic, facility accessibility, and land development intensity of each candidate site based on the preprocessed population travel origin and destination data, POI data, building data, and land use data, and to form a feature matrix for each candidate site.

[0126] The multi-objective optimization module is used to construct a multi-objective transportation station location model. Based on the feature matrix, candidate stations that meet the distance constraints are selected through a greedy search algorithm and then input into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station location model to obtain the optimal planning result.

[0127] The module division in this embodiment of the invention is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in the various embodiments of the invention can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0128] This embodiment also provides a computer device, which includes a processor and a memory. The memory is used to store a computer program (in this embodiment, the computer program includes a computing component and an iterative component, capable of model calculation and model updating). The computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, and is suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to realize the corresponding method flow or corresponding function. The processor described in this embodiment can be used in a multi-objective optimization method for urban rail transit station site selection.

[0129] This embodiment also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the multi-objective optimization method for urban rail transit station site selection in the above embodiment.

[0130] This embodiment also provides a computer program product, which includes a computer program that, when executed by a processor, implements the corresponding steps of the multi-objective optimization method for urban rail transit station site selection described in the above embodiment.

[0131] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention 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.

[0132] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. 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 illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0133] 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.

[0134] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment 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.

[0135] 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 it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A multi-objective optimization method for site selection of urban rail transit, characterized in that, Includes the following steps: Collect and preprocess urban road network data, population travel origin and destination data, POI data, building data, and land use data; Intersections of urban roads of level 3 and above are selected as candidate stations based on the preprocessed urban road network data, and the coordinates of the candidate stations are encoded. Based on the preprocessed population travel origin and destination data, POI data, building data and land use data, the service traffic, facility accessibility and land development intensity of each candidate site are quantified, and a feature matrix of each candidate site is formed; A multi-objective transportation station location selection model is constructed. Based on the feature matrix, candidate stations that meet the distance constraints are selected using a greedy search algorithm. The results are then input into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station location selection model and obtain the optimal planning result. The multi-objective transportation station location selection model includes a multi-objective optimization function and constraints; The multi-objective optimization function includes: The constraints include: wherein, is a service flow value for the candidate site , is a facility accessibility value for the candidate site , is a land use index for the candidate site , is a binary decision variable, is a number of selected candidate sites, is a set of candidate sites, is a distance between the candidate site and the candidate site , is a minimum distance; The steps involved in filtering candidate sites that meet distance constraints based on the feature matrix using a greedy search algorithm, and then inputting these candidate sites into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station location model to obtain the optimal planning result, specifically include: After screening candidate stations that meet the distance constraints using a greedy search algorithm, the results are input into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station location model, and the Pareto front solution set is output. Repeat the above steps until the maximum number of iterations is reached. Then, merge multiple Pareto front solution sets and perform non-dominated sorting to obtain the non-dominated solution as the global Pareto optimal solution, i.e., the optimal programming result. The steps of using a greedy search algorithm to filter candidate sites that meet distance constraints, inputting them into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station location model, and outputting the Pareto front solution set specifically include: Randomly select the candidate sites as the initial points. Based on the initial points and the feature matrix, use a greedy search algorithm to filter all candidate sites that meet the distance constraints as the input point set. The distance constraint is that the distance between two candidate sites is greater than or equal to a preset minimum distance. The input point set is input into NSGA-II to initialize the population, and individuals are encoded using binary encoding. Selected candidate sites within an individual are encoded as 1, and unselected candidate sites are encoded as 0. The fitness value of an individual is calculated according to the multi-objective optimization function. The population is non-dominated based on fitness values ​​to classify individuals into different Pareto front ranks, and the crowding degree of each individual in the same Pareto front rank is calculated. Two individuals are randomly selected from the current population using the binary tournament selection method. The individual with a lower Pareto front level and greater crowding is selected as the parent, and two-point crossover and bit-flip mutation are performed sequentially to obtain the offspring. The offspring and the current population are merged to obtain a new population. The new population is then sorted in a non-dominated manner. Individuals in the new population are divided into different Pareto front levels. After calculating the crowding degree of each individual in the same Pareto front level, the steps after randomly selecting two individuals are repeated until the maximum number of generations is reached and the Pareto front solution set is output.

2. The multi-objective optimization method for urban rail transit station site selection according to claim 1, characterized in that, The steps for quantifying the service traffic, facility accessibility, and land development intensity of each candidate site based on the preprocessed population travel origin-destination data, POI data, building data, and land use data specifically include: Based on the preprocessed origin and destination data of the population, the inbound and outbound passenger flow of the candidate stations are extracted, and the inbound and outbound passenger flow are added together to obtain the service passenger flow. Based on the preprocessed POI data, facility accessibility values ​​are calculated using an improved gravity model. The preprocessed land use data includes the total area of ​​the plot and the area ratio of land within the plot. The preprocessed building data includes the building footprint and building height. The land use mixed entropy value is calculated based on the area ratio of land within the plot. The plot ratio of the candidate site is calculated based on the total area of ​​the plot, the building footprint, and the building height. The land use mixed entropy value and the plot ratio are multiplied to obtain the land use index of the candidate plot.

3. The multi-objective optimization method for urban rail transit station site selection according to claim 1, characterized in that, The steps of collecting and preprocessing urban road network data, population travel origin and destination data, POI data, building data, and land use data specifically include: Based on the road classification of urban road network vector data, duplicate roads and roads below the urban third-level road classification in the urban road network data are removed; Based on the origin and destination data of the population, calculate the number of people leaving and entering the service area of ​​each pre-divided candidate station; The POI data is classified and weighted according to the entropy of the number of POI data in different categories. Calculate the building footprint and floor height based on building data, and calculate the total area of ​​the plot and the proportion of land area within the plot based on land use data.

4. A multi-objective optimization system for urban rail transit station site selection, based on the multi-objective optimization method for urban rail transit station site selection as described in any one of claims 1 to 3, characterized in that, include; The data acquisition and preprocessing module is used to collect urban road network data, population travel origin and destination data, POI data, building data, and land use data and perform preprocessing. The candidate site filtering module is used to filter the intersections of urban Class III roads and above as candidate sites based on the preprocessed urban road network data, and to encode the coordinates of the candidate sites. The feature extraction module is used to quantify the service traffic, facility accessibility, and land development intensity of each candidate site based on the preprocessed population travel origin and destination data, POI data, building data, and land use data, and to form a feature matrix for each candidate site. The multi-objective optimization module is used to construct a multi-objective transportation station location model. Based on the feature matrix, candidate stations that meet the distance constraints are selected through a greedy search algorithm and then input into NSGA-II to solve the multi-objective optimization function of the multi-objective transportation station location model to obtain the optimal planning result.

5. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable in the processor. When the processor executes the computer program, it implements the steps of the multi-objective optimization method for urban rail transit station site selection as described in any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the multi-objective optimization method for urban rail transit station site selection as described in any one of claims 1 to 3.

7. A computer program product, said computer program product comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the multi-objective optimization method for urban rail transit station site selection as described in any one of claims 1 to 3.

Images