Multi-source big data fused urban riding road network line selection method, system and device

Abstract

The invention relates to the field of urban planning and traffic engineering, and discloses a multi-source big data-fused urban riding road network line selection method, system and device, and the method comprises the steps: S1, riding supply measurement: carrying out the riding comfort scoring through a multi-modal big model, and carrying out the batch scoring after the scoring is calibrated through an iterative feedback mechanism; s2, riding demand measurement: carrying out track matching by adopting a hidden Markov model, counting the use frequency of road sections and calculating riding hotspots; and S3, road network line selection: calculating the comprehensive weight of each road section, establishing a total score minimum function of road network line selection, and generating an optimal scheme. According to the method, objective evaluation driven by data, accurate demand analysis, optimization of supply and demand matching and fusion utilization of multi-source data are achieved, scientificity and efficiency of urban riding space planning can be improved, green traffic development can be promoted, the urban environment can be improved, and powerful technical support is provided for bicycle-friendly city construction of various cities.

Description

Technical Field

[0001] The present invention relates to the fields of urban planning and traffic engineering, and specifically to a method, system and device for selecting a route for an urban cycling network by integrating multi-source big data. The method integrates street view image data, shared bicycle trajectory data and OD data, and realizes scientific planning and optimization of urban cycling space through multi-modal large model analysis, data mining and path optimization algorithms. Background Art

[0002] With the acceleration of urbanization and the popularization of the concept of green travel, bicycles are increasingly valued as an environmentally friendly and healthy mode of transportation. Many cities are actively promoting the construction of bicycle-friendly cities to improve urban traffic conditions, reduce environmental pollution, and improve the quality of life of residents. However, in the actual planning and construction of urban cycling spaces, there are still many challenges:

[0003] 1. Subjectivity of riding space quality assessment: Traditional riding space quality assessment methods often rely on manual surveys and expert scoring. This method is not only time-consuming and labor-intensive, but also easily affected by personal subjective factors, making it difficult to ensure the objectivity and consistency of the assessment results.

[0004] 2. Limitations of cycling demand analysis: Previous cycling demand analysis was mostly based on questionnaire surveys or small-scale GPS tracking data, which can hardly fully reflect the distribution and changing patterns of cycling demand in the entire city. The sample of this method is not representative enough and cannot provide strong support for large-scale urban cycling space planning.

[0005] 3. Ignoring the matching degree between supply and demand: Existing cycling route planning methods focus more on physical factors such as path length and slope, but ignore the matching degree between the quality of cycling space supply and actual demand. This may result in the planned cycling route looking reasonable on the map, but the actual utilization rate is not high.

[0006] 4. Insufficient data integration and utilization: Although the availability of various urban big data has greatly improved in recent years, it is still a challenge to effectively integrate and utilize these heterogeneous data sources and extract valuable information from them for cycling space planning.

[0007] 5. Scalability issues of planning methods: As cities expand in size and complexity, traditional planning methods often find it difficult to cope with the needs of large-scale data processing and complex scenario analysis, and lack good scalability.

[0008] 6. Insufficient adaptability to dynamic changes: A city is a dynamically changing system, and cycling demand and space utilization patterns are also constantly changing. Existing planning methods are mostly static analysis, which makes it difficult to reflect and adapt to these changes in a timely manner.

[0009] 7. Difficulty in multi-objective optimization: Cycling space planning involves multiple objectives, such as improving cycling experience, meeting cycling needs, and reducing construction costs. How to balance and optimize between these objectives is a complex issue.

[0010] In view of the above problems, there is an urgent need for an urban cycling space route selection method that can comprehensively consider the supply and demand of cycling space and is based on multi-source big data. This method should be able to objectively evaluate the quality of cycling space, accurately grasp the distribution of cycling demand, and on this basis, select the optimal route to support more scientific and efficient urban cycling space planning. Summary of the invention

[0011] The present invention aims to solve the technical problems existing in the prior art. To this end, the present invention provides a method for selecting a route for an urban cycling network that integrates multi-source big data. The method makes full use of multi-source big data such as street view image data, shared bicycle trajectory data, and OD data, and combines multimodal large models, machine learning algorithms, and path optimization technology to achieve a comprehensive analysis and optimization planning of urban cycling space, which can objectively evaluate the quality of cycling space, accurately grasp the distribution of cycling demand, and select the optimal route on this basis.

[0012] The technical solution adopted by the present invention to solve its technical problem is:

[0013] In a first aspect, a method for selecting a route for an urban cycling network by integrating multi-source big data is provided, comprising the following steps:

[0014] S1, cycling supply measurement:

[0015] Collect and preprocess street view image data;

[0016] Using a large multimodal model to score the riding comfort of street view images;

[0017] Calibrate model scores through an iterative feedback mechanism;

[0018] Batch score all street view images in the target area according to the calibrated scoring criteria;

[0019] S2, cycling demand measurement:

[0020] Collect shared bicycle trajectory data and OD data and perform cleaning and preprocessing;

[0021] Hidden Markov model is used for trajectory matching;

[0022] Count the frequency of road section use;

[0023] Use the HDBSCAN algorithm to identify cycling hotspots;

[0024] S3, road network selection:

[0025] Calculate the comprehensive weight of each road section according to the following formula:

[0026] W = w_l*(1 / l(e))+w_s*s(e)+w_f*f(e)+w_m*m(e); where w_l is the weight for controlling the path length, w_s is the weight for controlling the supply level, w_f is the weight for controlling the demand response, w_m is the weight for controlling the supply-demand matching, l(e) is the length of the section, s(e) is the supply score, f(e) is the demand intensity, and m(e) is the matching degree;

[0027] Establish the minimum total score function for road network selection: Min Z = Σ(W ij *F ij );W ij is the comprehensive weight of the road segment with i as the starting point and j as the end point, F ij is the usage frequency of the road segment with i as the starting point and j as the end point;

[0028] Generate the optimal solution and present it visually.

[0029] In some optional implementations, in step S1, model score calibration includes the following steps:

[0030] Input standard score street view image;

[0031] Compare the difference between the model score and the standard score;

[0032] Adjust the scoring criteria based on the difference;

[0033] When the difference in the scores of 10 consecutive images is within the preset range, it is confirmed that the scoring standard is stable.

[0034] In some optional embodiments, in step S2, the method for counting the frequency of use of a road section is as follows: based on the matched trajectory data, the number of times each road section is ridden is counted to obtain the riding frequency of each road section, which is used to reflect the intensity of demand for cycling behavior on the street. Cycling demand intensity = total number of rides / road section length. The total number of rides is the number of trajectories that pass through the road section according to the road section ID using the basic statistical method.

[0035] In some optional implementations, in step S3, the comprehensive weight calculation method for each road section is:

[0036] Using the acquired road network data, the road length l(e), supply score s(e) and demand intensity f(e) are calculated for each road section;

[0037] Calculate the matching degree between supply level and demand intensity of each road section;

[0038] Calculate the comprehensive weight of each road segment.

[0039] In some optional embodiments, the road section length l(e) is the actual length of the road section directly obtained from the road network data, in meters, after normalization; the supply score s(e) is the supply level obtained based on the street view score, in the range of [0,1], and a higher value indicates a better supply level; the demand intensity f(e) is the standardized usage frequency based on trajectory data statistics, in the range of [0,1], and a higher value indicates a stronger demand.

[0040] In some optional embodiments, the matching function is defined as: m(e)=1-|s(e)-f(e)| / max(s(e),f(e)); the value range is [0,1]; when the supply score s(e) and the demand intensity f(e) are completely matched, m(e)=1; the greater the difference between supply and demand, the smaller the m(e) value.

[0041] In some optional implementations, in the calculation formula of the comprehensive weight, to avoid the path being too long, the value range of w_l is [0.3, 0.6]; to ensure the comfort of the path, the value range of w_s is [0.1, 0.4]; to ensure service for actual needs, the value range of w_f is [0.1, 0.3]; the weight for controlling supply and demand matching gives priority to road sections with high supply and demand matching, and the value range of w_m is [0.05, 0.2].

[0042] In a second aspect, a system for realizing a method for selecting a route for an urban cycling network by integrating multi-source big data is provided, comprising:

[0043] Data collection and preprocessing module: responsible for collecting and cleaning multi-source big data such as shared bicycle trajectory data and street view images, and preprocessing them to provide accurate and usable data foundation for subsequent cycling supply and demand analysis;

[0044] Supply measurement analysis module: evaluates the comfort of the urban cycling environment and provides basic data for cycling route selection through street view image analysis and model calibration;

[0045] Demand measurement and analysis module: Analyze the usage data of shared bicycles, identify riding hotspots and the frequency of use of road sections, and understand the distribution of riding demand;

[0046] Demand measurement analysis module: integrates supply and demand data, generates the optimal cycling route plan through road section weight calculation and optimization algorithm, and displays it visually;

[0047] Road network selection and optimization module: Comprehensive supply and demand data, generate the optimal cycling route plan through road section weight calculation and optimization algorithm, and display it visually;

[0048] Visual display module: presents complex data analysis results and optimized cycling route plans in a graphical interface, allowing users to intuitively understand and effectively use this information to make decisions.

[0049] The data collection and preprocessing module in this system provides raw data for the entire system and processes the data to make it suitable for analysis; the supply measurement analysis module and the demand measurement analysis module analyze traffic conditions from the perspectives of supply and demand respectively; the road network line selection optimization module integrates the analysis results of supply and demand and proposes optimization plans; the visualization display module presents these complex analysis results to users in an intuitive way.

[0050] In a third aspect, a device is provided for implementing a method for selecting a route for an urban cycling network by integrating multi-source big data, including a computer-readable storage medium on which a computer program is stored, and the program implements each step of the method when executed by a processor.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] 1. The present invention realizes data-driven objective evaluation: by using street view images and multimodal large models, an objective and comprehensive evaluation of the quality of cycling spaces is achieved, overcoming the subjectivity and limitations of traditional manual evaluation methods;

[0053] 2. Accurate demand analysis: Based on large-scale shared bicycle data and combined with advanced spatiotemporal data mining technology, the spatial distribution characteristics and temporal variation patterns of riding demand are accurately grasped, providing a solid data foundation for planning and decision-making;

[0054] 3. Optimization of supply and demand matching: In the process of road network selection, supply and demand factors are comprehensively considered. Through multi-objective optimization methods, the supply and demand balance of cycling space planning is achieved, which improves the rationality and effectiveness of planning;

[0055] 4. Realized the integration and utilization of multi-source data: Innovatively integrated multi-source heterogeneous data such as street view images, shared bicycle tracks, and OD data, fully tapped the value of various types of data, and provided multi-dimensional information support for urban cycling space planning;

[0056] 5. Practicality of decision support: Through visual presentation and interactive tools, it provides intuitive and flexible decision support for planning decision makers, facilitating comparison, adjustment and optimization of plans;

[0057] 6. The implementation of the present invention can not only improve the scientificity and efficiency of urban cycling space planning, but also promote the development of green transportation, improve the urban environment, and enhance the quality of life of citizens. It has broad application prospects and can provide strong technical support for the construction of bicycle-friendly cities in various cities. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work, among which:

[0059] Figure 1 The overall flow chart of the urban cycling road network selection method provided by the present invention;

[0060] Figure 2 Schematic diagram of the process of evaluating cycling space quality in street images;

[0061] Figure 3 This is a flow chart for shared bicycle data processing;

[0062] Figure 4 This is a schematic diagram of the cycling hotspot clustering results based on the HDBSCAN algorithm;

[0063] Figure 5 This is the schematic diagram of the optimal cycling route that is finally generated. DETAILED DESCRIPTION

[0064] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0065] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in this field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0066] Based on a medium-scale test of cycling space planning, it is required to improve the cycling environment and increase the proportion of bicycle travel. It is hoped that the entire cycling network can be planned based on a data-driven approach. The following implementation plan is provided:

[0067] As attached Figure 1 As shown, a method for selecting a route for an urban cycling network integrating multi-source big data is provided, which is performed in the following steps:

[0068] S1, cycling supply measurement based on comprehensive score, specifically:

[0069] Get street view image data: Get street view images of the target area through the Baidu Map API interface. Each image contains metadata information such as pixel information, longitude and latitude information, shooting angle information, and timestamp.

[0070] Data preprocessing: preprocess the collected street view graphics, including image denoising and color correction. Specifically, use Gaussian filtering and other methods to remove image noise; adjust color balance according to the lighting conditions at the time of collection;

[0071] The cycling comfort rating of the street scene is evaluated based on the iterative feedback of the large model. The specific evaluation process is as follows:

[0072] Image preprocessing: The collected street view graphics are uniformly converted into 1024×768 resolution and Base64 encoding format to facilitate input into the large model and ensure that the image clarity meets the evaluation requirements;

[0073] Design the following scoring prompts:

[0074] ”'

[0075] How comfortable is the city scene in this picture for cycling? Please rate it from 0-1, where 0 means very uncomfortable and 1 means very comfortable. You must output the json as {"score":0.23,"reason":xxx}

[0076] ”'

[0077] The above standard prompt words are sent to a large model (such as GPT-4), and the large model is used to score the riding comfort of the street view image on a scale of 0-1, where 0 means very uncomfortable and 1 means very comfortable.

[0078] An iterative feedback mechanism is used to calibrate the model score, as shown in the attached Figure 2 As shown in FIG. 1 , an iterative feedback mechanism is specifically set up between two types of ratings: the rating given by volunteers based on street view images and the rating given by the large model based on the set prompt words.

[0079] Calculate the difference between the model score and the standard score; when the difference exceeds the upper limit of the preset interval, the feedback model score is too high and the scoring standard needs to be increased; when the difference is lower than the lower limit of the preset interval, the feedback model score is too low and the scoring standard needs to be lowered; when the difference is within the preset interval, confirm that the current score is appropriate and maintain the scoring standard; when the score differences of 10 consecutive images are all within the preset interval, confirm that the scoring standard is stable. In specific implementation, assuming that the preset interval is [-0.2, 0.2], when the difference is >0.2, feedback is given to the model: "The current scoring standard is too high, please lower the scoring standard appropriately", and re-scoring is performed; when the difference is <-0.2, feedback is given to the model: "The current scoring standard is too low, please increase the scoring standard appropriately", and re-scoring is performed; when the difference is within the interval of [-0.2, 0.2], feedback is given to the model: "The current scoring standard is appropriate, please maintain it", and continue to score the next image; continuously evaluate 10 images with standard scores, record the score difference of each image, and when the score difference of these 10 images are all within the interval of [-0.2, 0.2], it is confirmed that the scoring standard has reached stability.

[0080] Batch scoring: Keep the same cue template; follow the calibrated scoring criteria; score all street view images in the target area.

[0081] Monitoring of the scoring process: For every 100 images scored, a standard image is randomly inserted for testing. If the score difference of the test image exceeds the interval [-0.2, 0.2], the standard is recalibrated to ensure the continued stability of the scoring standard.

[0082] Comprehensive score calculation: Calculate the comprehensive cycling space quality score of each road section. The average of the image scores of all collected points on the road section is the score of the road section.

[0083] S2, the cycling demand measurement based on the total number of trajectories. Figure 3 As shown in the figure, the collected riding data is first cleaned to remove outliers, and then matched according to the road network to attach the trajectory to the road segment. Specifically:

[0084] Data collection: Cooperate with shared bicycle operating companies to collect anonymized trajectory data and OD (origin-destination) data within a certain period of time. The data should include information such as the ride order ID, the start and end time of the ride, the start and end location coordinates, the path points, and the longitude and latitude sequence of the trajectory points. When collecting data, the time range is 30 consecutive days; the spatial range is the study area and the surrounding 1 km buffer zone. After data collection, it is desensitized to ensure user privacy protection.

[0085] Data cleaning and preprocessing: Clean the collected raw data to remove outliers and erroneous data. Specifically: delete records with a cycling distance of less than 100 meters; delete records with a cycling distance of more than 20 kilometers; delete tracks that obviously deviate from the road network; delete records with severely discontinuous track points. Preprocessing includes time-space alignment, data format unification, etc.

[0086] Match the track to the road segment according to the road network, as follows:

[0087] Road network preprocessing: extract the road network in the study area, assign a unique ID to each road section, record the starting and ending coordinates of the road section, and establish a connection relationship table between road sections.

[0088] Trajectory matching: A map matching algorithm based on a hidden Markov model is used to map trajectory points to the most likely road segments. Specifically, for each trajectory point, the road segments within a range of 50 meters are searched, and the nearest 3-5 road segments are selected as candidates; the distance from the trajectory point to the road segment is recorded; the probability of the trajectory point to the candidate road segment is calculated, the connection relationship between the road segments is considered, and the Viterbi algorithm is used to determine the most likely path.

[0089] Road section usage frequency statistics: Based on the matched trajectory data, the number of times each road section is ridden is counted to obtain the riding frequency of each road section, which is used to reflect the intensity of cycling demand on the street. The specific road section cycling demand intensity is calculated as follows:

[0090] The number of trajectories that pass through the road section according to the road section ID is recorded as the total number of rides on the road section using the basic statistical method. The storage format of the statistical results is: road section ID, total number of rides, and road section length.

[0091] Calculate the amount of riding per unit length: riding demand intensity = total amount of riding / road section length.

[0092] Perform 0-1 normalization processing: Normalized value = (original value - minimum value) / (maximum value - minimum value).

[0093] Cycling hotspot identification: HDBSCAN (density-based hierarchical clustering algorithm) is used to cluster OD data and identify cycling hotspots, which will be used as the starting and ending points for subsequent line selection. In the cluster analysis based on the HDBSCAN algorithm, the minimum number of samples is 1000, the cluster radius is 500m, and the density threshold is determined according to the data distribution; the hotspot extraction process is as follows:

[0094] Data preparation: extract all OD point coordinates, build spatial index, and calculate point density distribution;

[0095] Cluster analysis: identify high-density areas, determine cluster boundaries, and merge adjacent clusters;

[0096] Hotspot identification: Calculate cluster centers, determine impact ranges, and evaluate hotspot intensity.

[0097] Output road section cycling demand characteristics and hotspot characteristics; the road section cycling demand characteristics include road section ID, total cycling volume, and standardized cycling demand intensity; the hotspot characteristics include hotspot ID and location, hotspot intensity level, service range radius, and surrounding road network connectivity.

[0098] According to the above hotspot identification method, the results are as shown in the attached Figure 4 As shown, the size of the dots represents the intensity level of the hotspot, and the larger the dot, the higher the intensity of the clustering.

[0099] S3, route selection for urban cycling, specifically:

[0100] Data preparation: Using the road network data (including road grades) obtained earlier, calculate the following three basic attribute values ​​for each road section:

[0101] Section length l(e): the actual length of the section directly obtained from the road network data, in meters, after normalization;

[0102] Supply score s(e): the supply level based on the street view score, ranging from [0,1], with higher values ​​indicating better supply levels;

[0103] Demand intensity f(e): The standardized usage frequency based on trajectory data statistics, with a value range of [0,1]. Higher values ​​indicate stronger demand.

[0104] Calculate the matching degree between the supply level and demand intensity of each road section. The matching function is defined as: m(e) = 1-|s(e)-f(e)| / max(s(e),f(e)). This function has the following characteristics:

[0105] The value range is [0,1]; when the supply score s(e) and the demand intensity f(e) are fully matched, m(e) = 1; the greater the difference between supply and demand, the smaller the m(e) value; max(s(e), f(e)) is used as the denominator for normalization, so that the matching degree of different road sections is comparable.

[0106] The comprehensive weight of each road section is calculated as follows:

[0107] W=w_l*(1 / l(e))+w_s*s(e)+w_f*f(e)+w_m*m(e)

[0108] Where: w_l is the weight for controlling the path length to avoid the path being too long, and its value range is [0.3, 0.6]; w_s is the weight for controlling the supply level to ensure the comfort of the path, and its value range is [0.1, 0.4]; w_f is the weight for controlling the demand response to ensure the service for the actual demand, and its value range is [0.1, 0.3]; w_m is the weight for controlling the supply and demand matching, giving priority to the road sections with high supply and demand matching, and its value range is [0.05, 0.2].

[0109] The overall optimization goal of road network selection is to minimize the total score, which is calculated using the following function:

[0110] Min Z=Σ(W ij *F ij )

[0111] Where W ij is the comprehensive weight of the road segment with i as the starting point and j as the end point, F ij is the usage frequency of the road segment with i as the starting point and j as the end point.

[0112] Based on the calculated comprehensive weight, the road network line selection is carried out. The results are shown in the attached Figure 5 As shown in the figure, the green dots represent cycling hotspots, the red road network represents the final selected road network result, and the numbers annotated on the road network represent the number of times the road section is used in the road network. The larger the number, the coarser the road network. The specific process is as follows:

[0113] Road section evaluation: calculate the comprehensive weight of each road section; establish the corresponding relationship between weight and road section; sort the road sections according to weight;

[0114] Road network generation: select road sections with higher weights as backbones; confirm the connectivity between key points; add necessary connecting sections;

[0115] Scheme evaluation: calculate the total weight of the road network; check the degree of supply and demand matching; verify network connectivity; calculate evaluation indicators, including total length, coverage, etc.

[0116] When problems are found in the solution, the following optimization strategies are adopted:

[0117] When the matching degree between supply and demand is generally low, increase the weight of w_m;

[0118] When the road network is too dispersed, increase the weight of w_l;

[0119] When the supply level is insufficient, increase the weight of w_s;

[0120] When demand response is insufficient, increase the weight of w_f.

[0121] Visualization and interaction: The final generated cycling route is visualized on the map and interactive exploration tools are provided.

[0122] The present invention utilizes the above steps and, through multiple rounds of iterative optimization, ultimately obtains a cycling road network solution with a high supply-demand matching degree and a reasonable network structure.

[0123] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for selecting urban cycling routes by integrating multi-source big data, characterized in that: The following steps are involved: S1, cycling supply measurement: Collect and preprocess street view image data; Using a large multimodal model to score the riding comfort of street view images; Calibrate model scores through an iterative feedback mechanism; Batch score all street view images in the target area according to the calibrated scoring criteria; S2, cycling demand measurement: Collect shared bicycle trajectory data and OD data and perform cleaning and preprocessing; Hidden Markov model is used for trajectory matching; Count the frequency of road section use; Use the HDBSCAN algorithm to identify cycling hotspots; S3, road network selection: Calculate the comprehensive weight of each road section according to the following formula: W = w_l*(1 / l(e))+w_s*s(e)+w_f*f(e)+w_m*m(e); where w_l is the weight for controlling the path length, w_s is the weight for controlling the supply level, w_f is the weight for controlling the demand response, w_m is the weight for controlling the supply-demand matching, l(e) is the length of the section, s(e) is the supply score, f(e) is the demand intensity, and m(e) is the matching degree; Establish the minimum total score function for road network selection: Min Z = Σ(W ij *F ij );W ij is the comprehensive weight of the road segment with i as the starting point and j as the end point, F ij is the usage frequency of the road segment with i as the starting point and j as the end point; Generate the optimal solution and present it visually.

2. The method according to claim 1, characterized in that In step S1, model scoring calibration includes the following steps: Input standard score street view image; Compare the difference between the model score and the standard score; Adjust the scoring criteria based on the difference; When the difference in the scores of 10 consecutive images is within the preset range, it is confirmed that the scoring standard is stable.

3. The method according to claim 1, characterized in that In step S2, the method for counting the frequency of use of the road section is as follows: based on the matched trajectory data, the number of times each road section is ridden is counted to obtain the riding frequency of each road section, which is used to reflect the intensity of the demand for riding behavior on the street. The riding demand intensity = total riding volume / road section length. The total riding volume is the number of trajectories that pass through the road section according to the road section ID using the basic statistical method.

4. The method according to claim 1, characterized in that: In step S3, the comprehensive weight calculation method for each road section is: Using the acquired road network data, the road length l(e), supply score s(e) and demand intensity f(e) are calculated for each road section; Calculate the matching degree between supply level and demand intensity of each road section; Calculate the comprehensive weight of each road segment.

5. The method according to claim 4, characterized in that The road section length l(e) is the actual length of the road section directly obtained from the road network data, in meters, after normalization; the supply score s(e) is the supply level obtained based on the street view score, with a value range of [0,1], and a higher value indicates a better supply level; the demand intensity f(e) is the standardized usage frequency based on trajectory data statistics, with a value range of [0,1], and a higher value indicates a stronger demand.

6. The method according to claim 4, characterized in that The matching function is defined as: m(e)=1-|s(e)-f(e)| / max(s(e),f(e)); the value range is [0,1]; when the supply score s(e) and the demand intensity f(e) are fully matched, m(e)=1; the greater the difference between supply and demand, the smaller the m(e) value.

7. The method according to claim 1, characterized in that In the calculation formula of the comprehensive weight, in order to avoid the path being too long, the value range of w_l is [0.3, 0.6]; to ensure the comfort of the path, the value range of w_s is [0.1, 0.4]; to ensure service for actual needs, the value range of w_f is [0.1, 0.3]; the weight for controlling supply and demand matching gives priority to sections with high supply and demand matching, and the value range of w_m is [0.05, 0.2].

8. A system for implementing the method according to any one of claims 1 to 7, characterized in that: include: Data collection and preprocessing module: responsible for collecting and cleaning multi-source big data such as shared bicycle trajectory data and street view images, and preprocessing them to provide accurate and usable data foundation for subsequent cycling supply and demand analysis; Supply measurement analysis module: evaluates the comfort of the urban cycling environment and provides basic data for cycling route selection through street view image analysis and model calibration; Demand measurement analysis module: Analyze the usage data of shared bicycles, identify riding hotspots and the frequency of use of road sections, and understand the distribution of riding demand; Demand measurement analysis module: integrates supply and demand data, generates the optimal cycling route plan through road section weight calculation and optimization algorithm, and displays it visually; Road network route selection optimization module: Comprehensive supply and demand data, generate the optimal cycling route plan through road section weight calculation and optimization algorithm, and display it visually; Visual display module: presents complex data analysis results and optimized cycling route plans in a graphical interface, allowing users to intuitively understand and effectively use this information to make decisions.

9. A device for implementing the method according to any one of claims 1 to 7, characterized in that: The invention comprises a computer-readable storage medium, wherein a computer program is stored on the storage medium, and when the program is executed by a processor, each step of the method is implemented.

Images