An urban path network supplement point optimization method based on improved dijkstra algorithm

Abstract

This invention relates to the field of digital urban planning and intelligent path optimization technology, and discloses a method for optimizing urban path network replacement points based on an improved Dijkstra algorithm. First, a county-level facility-road network coupled graph model is constructed, and facility load levels and population weights are labeled. The standard Dijkstra algorithm is used to calculate the initial coverage rate and load rate. Based on the load level and coverage blind spots, a load-blind spot coupled gravitational field model is constructed, and AI adaptive weights are used to generate candidate replacement point heatmaps. Coverage gain, blind spot attraction, and load diversion factors are introduced to construct a multi-dimensional cost function for the improved Dijkstra algorithm, and the coefficients are dynamically adjusted using AI adaptive weights. The comprehensive cost is calculated by traversing the candidate points, and the point with the minimum comprehensive cost is selected as the optimal replacement point. The process is iterated until a termination condition is met, and then the solution is output. This solves the problem of not being able to automatically locate the optimal replacement point position that balances coverage improvement and load diversion.

Description

Technical Field

[0001] This invention relates to the field of digital urban planning and intelligent path optimization technology, and in particular to a method for optimizing urban path network point supplementation based on an improved Dijkstra algorithm. Background Technology

[0002] In the fields of urban planning and intelligent path optimization, improving road network coverage and the rational layout of public service facilities (such as medical points, bus stops, and entrances / exits to pedestrian and cycling systems) are crucial aspects of urban renewal and infrastructure development. Existing path planning methods mostly employ the standard Dijkstra algorithm, which uses the shortest path length as its sole optimization objective, calculating the shortest path from the starting point to each node through greedy point selection and relaxation updates. However, the standard Dijkstra algorithm only focuses on travel distance or time cost, failing to assess the service coverage of the road network within the urban space, identify coverage blind spots, or consider the impact of the load on facility nodes (such as the number of people served by medical points and the passenger flow intensity of bus stops) on the effectiveness of point replacement. Therefore, relying solely on the standard Dijkstra algorithm is insufficient to meet the planning needs for improving the accessibility of public service facilities and optimizing the road network in county-level areas.

[0003] In practical engineering applications, the addition of county-level public service facility nodes (such as bus stops, medical service points, and pedestrian entrances / exits) often relies on manual experience and judgment, lacking quantitative basis. This traditional deployment method is prone to problems such as unreasonable location of addition points, limited coverage improvement, unbalanced facility load, and waste of resources, failing to simultaneously achieve the dual goals of "filling coverage blind spots" and "distributing the load on existing facilities." Furthermore, existing algorithms lack the ability to quantitatively evaluate the linkage between "addition points—coverage improvement—load balancing," and cannot automatically locate the optimal addition points that can significantly improve coverage while effectively distributing the pressure on high-load facilities, making it difficult to achieve simultaneous improvement in road network coverage and facility service efficiency.

[0004] In summary, existing technologies cannot accurately identify and weightedly calculate blind spots by combining factors such as community distribution, facility load, and population distribution at the county scale. There is a lack of a path network supplementation optimization method that can automatically identify coverage blind spots and high-load facilities, and quantitatively evaluate the dual impact of different supplementation locations on coverage and load balance. Therefore, there is an urgent need to develop an optimization method to overcome the shortcomings of traditional algorithms, such as a single optimization objective, neglect of load levels, and reliance on human experience. This method would provide scientific and quantitative technical support for scenarios such as the supplementation of county-level public service facilities, optimization of slow-traffic systems, and renewal of old urban areas. Summary of the Invention

[0005] The purpose of this invention is to provide an optimization method for urban path network replacement points based on an improved Dijkstra algorithm, which solves the problem that existing technologies cannot automatically locate the optimal replacement point position in urban path networks that balances coverage improvement and load distribution.

[0006] To achieve the above objectives, this invention provides a method for optimizing urban path network point supplementation based on an improved Dijkstra algorithm, comprising the following steps:

[0007] S1: Construct a county-level infrastructure-road network coupled graph model. The node set of the county-level infrastructure-road network coupled graph model includes public service facility nodes, residential community nodes, and road intersection nodes. Mark the real-time load level of each public service facility node and mark the population size weight of each residential community node.

[0008] S2: The initial reachability and facility load status are calculated using the standard Dijkstra algorithm to obtain the initial coverage and the initial load rate of each facility node;

[0009] S3: Based on the facility load level and coverage blind spot in the county facility-road network coupling diagram model, construct a load-blind spot coupled gravitational field model, and use the first AI adaptive weight decision model to dynamically adjust the gravitational field weight coefficient to generate candidate supplementary point thermal zones.

[0010] S4: Introduce coverage gain factor, blind zone coverage reward factor and load balancing factor to construct a multi-dimensional cost function for the improved Dijkstra algorithm, and use the second AI adaptive weight decision model to dynamically adjust the weight coefficients of each factor in the cost function.

[0011] S5: Traverse each candidate point in the candidate replacement point heat zone, temporarily add the candidate point to the county infrastructure-road network coupling graph model, run the improved Dijkstra algorithm to calculate the comprehensive cost of each candidate point, and select the candidate point with the minimum comprehensive cost as the optimal replacement point;

[0012] S6: Update the optimal replacement point to the county infrastructure-road network coupled graph model, iteratively execute steps S2 to S5 until the iteration termination condition is met, and output the replacement point optimization scheme.

[0013] Specifically, step S1 includes:

[0014] The real-time load level is the ratio of the resident population covered by the public service facility node to its rated service capacity.

[0015] Specifically, step S3 includes:

[0016] The calculation formula for the load-blind zone coupled gravitational field model is as follows:

[0017]

[0018] in, Points Coverage blind spot density in the area For point The average load level of surrounding facility nodes For point The size of the resident population within the service area, , , This is the gravitational field weighting coefficient.

[0019] Specifically, the method of dynamically adjusting the gravitational field weight coefficients using the first AI adaptive weight decision model includes:

[0020] Input the current coverage, load balancing, population density, and scene type features. Perform coverage-driven correction, load balancing-driven correction, and population density-driven correction, and normalize the corrected weights to meet the conditions. .

[0021] Specifically, step S4 includes:

[0022] The multidimensional cost function of the improved Dijkstra algorithm is:

[0023]

[0024] in, Supplementing candidate points The average shortest path time cost to surrounding residential areas and road nodes. For candidate points The increase in coverage after the construction of additional facilities Candidate points The load-blind zone coupling gravity value, For candidate points The extent to which the load on high-load facilities is reduced after the construction of additional facilities. This is the cost weighting coefficient.

[0025] Specifically, the second AI adaptive weight decision model is used to dynamically adjust the weight coefficients of each factor in the cost function, including:

[0026] Input the current coverage gap, load imbalance, population concentration, and scenario type characteristics. Perform coverage-driven correction, load balancing-driven correction, and population density-driven correction, and normalize the corrected weights to meet the conditions. .

[0027] After generating the candidate thermal regions in step S3, the process also includes:

[0028] A set of candidate replacement points is generated within the candidate replacement point heat zone. The set of candidate replacement points includes locations where facilities can be deployed, such as the middle section of a road, road intersections, and entrances / exits of residential areas.

[0029] The iteration termination condition in step S6 is:

[0030] The county's facility coverage rate reaches a preset threshold and the facility load balance reaches a preset threshold. Before each iteration, steps S2 to S5 are executed, and the load variation coefficient of each public service facility node is recalculated based on the updated county facility-road network coupling graph model to quantify the facility load balance.

[0031] The method further includes:

[0032] An online interactive visualization platform based on OSM maps is built to display the coverage status, load level, path distribution and coverage blind spots before and after point addition, and generate interactive result files.

[0033] In step S5, after selecting the candidate point with the minimum comprehensive cost as the optimal replacement point, the following steps are also included:

[0034] Calculate the overall gain value of the optimal compensation point: ,in, For the increase in coverage, To optimize the load amplitude, , These are the weighting coefficients.

[0035] This invention discloses a method for optimizing urban path network nodes based on an improved Dijkstra algorithm. First, it constructs a county-level facility-road network coupled graph model including public service facility nodes, residential community nodes, and road intersection nodes, labeling facility nodes with real-time load levels and residential community nodes with population size weights. Second, it uses the standard Dijkstra algorithm to calculate the initial reachability and facility load status, obtaining the initial coverage rate and the initial load rate of each facility node. Then, based on the facility load levels and coverage blind spots in the graph model, it constructs a load-blind spot coupled gravitational field model, and dynamically adjusts the gravitational field using a first AI adaptive weight decision model. Weighting coefficients are used to generate candidate replacement point heatmaps. Then, coverage gain factors, blind zone coverage reward factors, and load diversion factors are introduced to construct a multi-dimensional cost function for the improved Dijkstra algorithm. A second AI adaptive weight decision model is used to dynamically adjust the weighting coefficients of each factor in the cost function. Next, each candidate point in the candidate replacement point heatmap is traversed, temporarily added to the graph model, and the improved Dijkstra algorithm is run to calculate the comprehensive cost of each candidate point. The candidate point with the minimum comprehensive cost is selected as the optimal replacement point. Finally, the optimal replacement point is updated to the graph model, and the above steps are iteratively executed until the iteration termination condition is met, outputting the replacement point optimization scheme. This effectively solves the problem that existing technologies cannot automatically locate the optimal replacement point position in urban path networks that balances coverage improvement and load diversion, achieving the dual optimization goals of rapid coverage improvement and effective load diversion after single-point replacement. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0037] Figure 1 This is a flowchart of the urban path network point supplementation optimization method based on the improved Dijkstra algorithm of the present invention.

[0038] Figure 2 This is the logic diagram for calculating the cost function of the improved Dijkstra algorithm of the present invention.

[0039] Figure 3 This is a schematic diagram comparing the road network coverage before and after the supplementary points of the present invention.

[0040] Figure 4 This is a flowchart illustrating the steps of the urban path network point supplementation optimization method based on the improved Dijkstra algorithm of the present invention. Detailed Implementation

[0041] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, but should not be construed as limiting the present invention.

[0042] Please refer to Figures 1 to 4 This invention provides a method for optimizing urban path network point supplementation based on an improved Dijkstra algorithm, comprising the following steps:

[0043] S1: Construct a county-level infrastructure-road network coupled graph model. The node set of the county-level infrastructure-road network coupled graph model includes public service facility nodes, residential community nodes, and road intersection nodes. Mark the real-time load level of each public service facility node and mark the population size weight of each residential community node.

[0044] Specifically, city roads are abstracted as a topological graph G=(V, E):

[0045] Node set V: includes three categories: public service facility nodes (medical points, bus stops, etc.), residential community nodes, and road intersection nodes. The real-time load level is marked for each facility node. The real-time load level specifically refers to the ratio of the resident population size covered by the facility node to its rated service capacity, such as the number of people served by the medical point and the passenger flow intensity of the bus stop. The population size weight is marked for the community nodes.

[0046] Edge set E: Corresponds to road segments in the county, and is assigned attributes such as road segment length, walking or public transportation travel time, and travel cost;

[0047] Based on OSM maps, geospatial mapping of graph models can be completed, and online interactive visualization platforms can be built on these graph models to intuitively display node distribution, coverage status, and load levels.

[0048] S2: The initial reachability and facility load status are calculated using the standard Dijkstra algorithm to obtain the initial coverage and the initial load rate of each facility node;

[0049] Specifically, this embodiment takes the optimization of county-level medical service facilities as an example. The initial state is quantified using the standard Dijkstra algorithm to provide basic input for subsequent optimization. The core calculation process is as follows:

[0050] 2.1 Core logic of the standard Dijkstra algorithm:

[0051] 1. Initialization: Starting Point shortest path The remaining nodes Mark all nodes as unvisited;

[0052] 2. Greedy selection: At each step, select the node with the smallest shortest path from the unvisited nodes. ,Right now ,mark Visited;

[0053] 3. Relaxed Update: For All adjacent nodes Perform relaxation operations to update the shortest path:

[0054]

[0055] in For nodes arrive Combined travel time for walking and public transportation;

[0056] 4. Iteration Termination: Repeat the point selection and relaxation operations until all nodes have been visited, then output all nodes back to the starting point. The shortest path time is used to generate an initial reachability matrix (example matrix below):

[0057]

[0058] 2.2 Initial Coverage State Calculation:

[0059] The achievable threshold is 15 minutes. Coverage determination for 49 cell nodes in a certain county based on the reachability matrix:

[0060] There are 13 uncovered cells and 36 covered cells, with an initial coverage rate of [missing information]. The northern section of Jianmen Avenue was identified as the core coverage blind area. This initial coverage rate will serve as the benchmark value for identifying coverage blind areas in step S3 and calculating coverage gain in step S4.

[0061] 2.3 Quantification of facility load rate:

[0062] Based on the quantification of the load of each medical facility according to its accessibility, the load rate formula is:

[0063]

[0064] in For facilities Population size covered This is the rated service capacity. The load rate of the First People's Hospital of a certain county has been calculated. The load level far exceeds the 100% threshold, classifying it as a high-load facility node within the county. This high-load facility node will serve as the core object for the subsequent gravitational field construction in step S3 and the load diversion assessment in step S4.

[0065] S3: Based on the facility load level and coverage blind spot in the county facility-road network coupling diagram model, construct a load-blind spot coupled gravitational field model, and use the first AI adaptive weight decision model to dynamically adjust the gravitational field weight coefficient to generate candidate supplementary point thermal zones.

[0066] Specifically, following the initial coverage blind spots and high-load facility nodes calculated in step S2, a load-blind spot coupled gravitational field model is constructed based on the dual dimensions of facility load level and coverage blind spots in the graph model. This model is used to initially screen high-value candidate replacement point areas. It should be noted that this gravitational field model is only used for preliminary location of candidate areas and is not directly used as the basis for final point selection. The final optimal replacement point location is determined by subsequent evaluation of the coverage and load diversion effect of each candidate point using the improved Dijkstra algorithm.

[0067] Abandoning the traditional screening logic that focuses solely on blind spot density, this approach combines the facility load level and coverage blind spot dimensions of a graph model to locate candidate areas.

[0068] Step 1: Filter out cells with coverage blind spots and match the load status of surrounding facility nodes. Prioritize uncovered cells within the service coverage range of high-load facility nodes as the core optimization area.

[0069] Step 2: Construct a load-blind zone coupled gravitational field model. The formula for calculating the gravitational value is as follows:

[0070]

[0071] in, Points The density of coverage blind spots in the area, i.e., the percentage of uncovered cells; Points The average load level of surrounding facility nodes, i.e., the proportion of high-load facilities or the overload rate; Points The size of the resident population within the service area. This is the gravitational field weighting coefficient.

[0072] against Adaptive weight coefficients: This invention constructs the first AI adaptive weight decision model, realizing automated and intelligent dynamic adjustment of weights, completely replacing manual intervention. The decision logic of this model is as follows:

[0073] (1) Input features of the AI ​​weighted decision model:

[0074] Taking the county-level medical facility planning scenario as an example, the model input includes four core features, comprehensively covering the planning objectives and the current operational status:

[0075] Scene type characteristics Differentiate planning scenarios for different public service facilities such as medical care, public transportation, pedestrian and bicycle paths, and emergency response, and preset benchmark weights for each scenario;

[0076] Coverage features The current coverage rate of public service facilities within a 15-minute reach in the county, quantifying the coverage gap;

[0077] Load balancing characteristics The load variation coefficient of county-level facility nodes quantifies the degree of facility load imbalance;

[0078] Population density characteristics : The spatial distribution concentration of the county's population, quantifying the scale of service demand.

[0079] (2) Weight adaptive adjustment rule:

[0080] Based on input features, a multi-dimensional weight correction logic is constructed to achieve dynamic weight adaptation:

[0081] Coverage-driven revision: based on county-level planning target coverage (e.g., medical scenarios) Using as a benchmark, calculate the coverage gap. ;

[0082] when (Low Coverage Phase): Triggers high coverage priority correction, Multiply by the correction factor At the same time Multiply by respectively Prioritize increasing the weight of blind spot coverage;

[0083] when (Mid-coverage phase): Triggers balance adjustment. Multiply , Multiply Version 1.05 balances coverage and load;

[0084] when (High Coverage Phase): Trigger load priority correction, Multiply by the correction factor 1.2, and at the same time Multiply Prioritize increasing the load balancing weight.

[0085] Load balancing-driven correction: based on load variability threshold Calculate the load imbalance degree based on the baseline. ;

[0086] when (High Imbalance Phase): Triggers emergency load correction, Multiply by the correction factor At the same time Multiply Strengthen load balancing weights;

[0087] when (Medium Imbalance Phase): Triggers a moderate load correction. Multiply ;

[0088] when (Equilibrium Phase): Maintaining Benchmark weights.

[0089] Population density-driven correction: based on average population density in counties Based on this, calculate the population concentration at the location. :

[0090] when (High-density areas): Trigger population priority adjustment, which will... Multiply by the correction factor Prioritize the replenishment of points in densely populated areas;

[0091] when (Low-density areas): Trigger population reduction adjustment, which will... Multiply 0.9, reduce the priority of patching points in low-density areas.

[0092] (3) Weight normalization constraint:

[0093] To ensure the dimensional consistency and rationality of the gravity calculation, all corrected weights must be normalized to meet the following requirements. The normalization formula is: in, As the baseline weight for the scenario, Correction coefficients for the corresponding dimensions, The final adaptive weights are used for the output.

[0094] (4) Example of AI weight decision-making in a certain county:

[0095] Taking the planning of county-level medical facilities in a certain county as an example, the input features are as follows:

[0096] Scene type Medical facility reconstruction, benchmark weighting ;

[0097] Current coverage Target coverage (Low coverage phase);

[0098] Load variation coefficient threshold (High Imbalance Stage);

[0099] Population concentration at the bridgehead of Xiucheng Bridge (High-density area).

[0100] Perform weight adjustment:

[0101] Coverage Correction: ;

[0102] Load correction: ;

[0103] Population Correction ;

[0104] Normalization process:

[0105]

[0106] The final output is the adaptive weights for a certain county scene. Substituting the gravitational field formula, the gravitational value of the Xiucheng Bridgehead area was calculated, verifying that it is a candidate area with the highest gravitational value in the county, and providing accurate input for subsequent improvements to the Dijkstra algorithm.

[0107] The third step is to generate candidate thermal regions for point replacement based on the gravitational field model. Regions with higher gravity values ​​are candidate regions with higher priority for point replacement, providing a set of points to be evaluated for subsequent improvements to the Dijkstra algorithm.

[0108] S4: Introduce coverage gain factor, blind zone coverage reward factor and load balancing factor to construct a multi-dimensional cost function for the improved Dijkstra algorithm, and use the second AI adaptive weight decision model to dynamically adjust the weight coefficients of each factor in the cost function.

[0109] Specifically, based on the candidate replacement point heatmap and gravity values ​​generated by S3, and building upon the traditional Dijkstra algorithm which only considers path distance as the sole cost, three scenario-based cost factors are added: coverage gain, blind zone coverage reward, and load balancing. This constructs a multi-dimensional improved cost function, achieving precise optimization of medical facility replacement points. The improved cost function is as follows:

[0110]

[0111] For the scenario of supplementing and rebuilding medical facilities in a certain county, the factors of each dimension of the improved cost function are defined as follows:

[0112] The average shortest path time cost from candidate point n to surrounding residential areas and road nodes reflects the accessibility of the point.

[0113] At candidate point The increase in the coverage rate of the 15-minute medical service circle in the entire county after the reconstruction of medical facilities is used to quantify the coverage gain effect of the reconstruction.

[0114] Candidate points The load-blind zone coupling gravity value, calculated by the gravitational field model mentioned above, reflects the coverage potential of the supplementary point on the core blind zone of the county.

[0115] After constructing additional medical facilities at candidate point n, the load reduction of a county's core high-load medical facilities (such as the county's First People's Hospital) is quantified to assess the load diversion effect of the additional facilities.

[0116] To unify the dimensions and avoid unit conflicts, all indicators are normalized to the [0, 1] interval, and each dimension indicator is converted into a unitless relative value before being weighted and summed to ensure the scientific and reasonable nature of the cost calculation.

[0117] Adaptive weight coefficients in the cost function This invention constructs a second AI adaptive weight decision model to achieve fully automated dynamic adjustment of weights, completely replacing manual intervention. The decision logic is explained in detail below using a county-level scenario:

[0118] 4.1 Second AI Adaptive Weight Decision Model (Adapted to a County Scenario):

[0119] (1) Model input features (measured data from a certain county):

[0120] Taking the reconstruction of county-level medical facilities as the core scenario, the model input includes four core features, comprehensively covering the county's planning goals and current operational status:

[0121] Scenario type characteristics: Medical facility reconstruction scenario, with preset baseline weights. ;

[0122] Coverage characteristics: The initial 15-minute medical service circle coverage rate in a certain county was 74%, with a planned target coverage rate of 98%, indicating a coverage gap. It is currently in a low coverage phase;

[0123] Load balancing characteristics: The initial load rate of a core medical facility (the county's First People's Hospital) in a certain county is 125%, the coefficient of variation of the county's facility load is 0.32, and the load imbalance degree is... It is in a stage of high imbalance;

[0124] Population density characteristics: The northern section of Jianmen Avenue in a certain county is the core residential area of ​​the county, with a population concentration of 1.8, which is much higher than the county average.

[0125] (2) Multi-dimensional weight adaptive correction rule:

[0126] Based on measured characteristics of a certain county, multi-dimensional weight adjustments are performed to achieve dynamic adaptation of weights to county planning objectives:

[0127] Coverage-driven correction: targeting a 24% coverage gap in a certain county ( (During the low coverage phase), a high coverage priority correction is triggered, which adjusts the coverage gain weight. Blind zone gravity weight Multiply by correction factors of 1.2 and 1.1 respectively, and simultaneously adjust the distance weight. Multiply by 0.95 to prioritize increasing the coverage contribution of supplementary points;

[0128] Load balancing-driven correction: targeting a load imbalance of 0.12 in a certain county ( During a high imbalance phase, an emergency load correction is triggered, reducing the load weight. Multiply by a correction factor of 1.25 to enhance the load balancing value of the supplementary points;

[0129] Population density-driven correction: For the Jianmen Avenue North Section area, a population concentration of 1.8 was applied. High-density areas), triggering population priority adjustments, will The weights are multiplied by 1.05 to increase the priority of adding points in high-density residential areas.

[0130] (3) Weight normalization and final output:

[0131] To ensure the rationality and dimensional consistency of the weights, the corrected weights are normalized to satisfy the following conditions: The normalization formula is:

[0132]

[0133] in As the baseline weight for the scenario, Correction coefficients for the corresponding dimensions, This represents the final adaptive weights. Combining the correction and normalization calculations with the scenario of a specific county, the optimal adaptive weights for the medical infrastructure reconstruction scenario in that county are finally output: This weighting perfectly aligns with the county's core planning objective of "prioritizing the improvement of medical coverage while taking into account the balanced load of facilities."

[0134] 4.2 Improved Dijkstra's algorithm for practical application in a county-level scenario:

[0135] The candidate site is located at the bridgehead of the Xiucheng Bridge on the northern section of Jianmen Avenue in a certain county. Using AI adaptive weights as the core computational object, the improved cost function is calculated in a scenario-based manner.

[0136] (1) Normalized values ​​of indicators for each dimension:

[0137] The candidate point is located next to the main road of Jianmen Avenue. The average shortest path time to the surrounding communities is 8 minutes, and the normalized value is 0.38.

[0138] After the reconstruction, five new communities that can be reached within 15 minutes will be added, increasing the coverage rate from 74% to 84%, with a normalized value of 0.91 for the coverage rate increase.

[0139] Based on the gravitational field model mentioned above, the gravitational value of this plot is 0.7995, and the normalized value is 0.80.

[0140] After the reconstruction, it can divert 8% of the patient volume of the First People's Hospital of a certain county, and the normalized value of the load reduction is 0.85.

[0141] (2) Improved cost function calculation:

[0142] Substitute the normalized values ​​of each indicator and the adaptive weights into the improved cost function to calculate the candidate points. The overall cost:

[0143]

[0144] (3) Comparison of candidate points and determination of the optimal replacement point:

[0145] Compared with the comprehensive costs of the other three candidate sites in a certain county (candidate site A: 0.213, candidate site B: 0.187, candidate site C: 0.152), the site at the bridgehead of Xiucheng Bridge has the lowest comprehensive cost. Therefore, it was determined to be the optimal location for the first round of medical reconstruction in the county. After reconstruction, the coverage rate can be increased from 74% to 84%, and the load rate of the core hospital can be reduced from 125% to 116%, which fully matches the county's planning goals.

[0146] 4.3 Algorithm Iteration Optimization and Full-Process Closed Loop:

[0147] After completing the first round of supplementary data, this invention automatically triggers a second round of decision-making using a second AI adaptive weighted decision model based on updated data on the county's medical coverage status and facility load. The weight coefficients are dynamically adjusted, and the improved Dijkstra algorithm is re-run to complete the second round of optimization simulation. By adding a new bus route "Xiucheng Bridge - People's Court - County People's Government - Tax Bureau - Shuren Jiayuan - Baicaoyuan Community - County First People's Hospital - Mingzhu Shangcheng", the path cost is further optimized. Ultimately, the coverage rate of the 15-minute medical service circle in the county is increased to 98%, the load rate of the core hospital is reduced to 104%, and the load balance reaches 93%, completing the closed loop of the entire process of optimizing county-level medical facilities.

[0148] S5: Traverse each candidate point in the candidate replacement point heat zone, temporarily add the candidate point to the county infrastructure-road network coupling graph model, run the improved Dijkstra algorithm to calculate the comprehensive cost of each candidate point, and select the candidate point with the minimum comprehensive cost as the optimal replacement point;

[0149] Specifically, building upon the improved Dijkstra algorithm constructed in S4 and the candidate point thermal region generated in S3, a set of candidate points (such as locations where facilities / nodes can be deployed, like the middle section of a road, intersections, and entrances / exits of residential areas) is generated within the load-blind zone coupled gravitational field thermal region. The following operations are performed on each candidate point:

[0150] (1) Temporarily add candidate points to the graph model topology and update the road network connectivity;

[0151] (2) Run the improved Dijkstra algorithm to recalculate the reachability, coverage, and load of each facility node in the county;

[0152] (3) Quantitative calculation of the overall gain value after point supplementation: ,in For the increase in coverage, To optimize the load amplitude, , These are the weighting coefficients;

[0153] (4) Select the location with the largest comprehensive gain value as the optimal supplementary point to ensure that the coverage can be greatly improved by adding a single point, while effectively diverting the service pressure of high-load facilities.

[0154] S6: Update the optimal replacement point to the county infrastructure-road network coupled graph model, iteratively execute steps S2 to S5 until the iteration termination condition is met, and output the replacement point optimization scheme.

[0155] Specifically, update the optimal replacement point to the graph model, repeat steps S2-S5, iteratively optimize the replacement point scheme until the county's facility coverage rate reaches the planning threshold (e.g., ≥95%) and the facility load balance (e.g., ≥95%) meets the requirements, then stop the iteration;

[0156] An online interactive platform is built based on OSM maps to visualize the coverage status, load level, path distribution and coverage blind spots before and after point addition, and generates an interactive 15minswalking_map.html output file to support planning decisions.

[0157] First, by combining the load-blind zone coupled gravitational field model with the multi-dimensional cost function of the improved Dijkstra algorithm, a single point addition can quickly close the coverage blind zone, resulting in a coverage improvement far exceeding that of traditional empirical point placement. Second, the improved Dijkstra algorithm simultaneously considers the triple objectives of shortest path, coverage gain, and load optimization, filling the gap in traditional algorithms that do not assess facility load, and perfectly adapting to the accessibility optimization needs of county-level public service facilities (such as medical care and public transportation). Third, this invention eliminates the need for subjective human judgment, automatically completing graph model construction, load assessment, blind zone identification, and optimal point addition. The entire positioning process is based on sufficient quantitative data, and its planning efficiency far exceeds that of manual experience-based point placement. Furthermore, this invention utilizes an online, interactive visualization platform built on OSM maps to intuitively display coverage status and load distribution before and after point placement. It can be directly applied to practical engineering scenarios such as county-level urban health checks, public service facility upgrades, slow-traffic system optimization, and old residential area renovation, demonstrating strong practicality. Finally, this invention boasts stable computation, supports deep integration with GIS, CIM (City Information Modeling), and multi-source population and traffic data, and is compatible with county or urban road networks of different scales, flexibly adapting to the optimization needs of various types of public service facilities. Therefore, this invention effectively solves the problem of existing technologies being unable to automatically locate the optimal point placement location that balances coverage improvement and load distribution, achieving the dual optimization goals of road network coverage and facility load balancing.

[0158] The above-disclosed embodiments are merely one or more preferred embodiments of this application and should not be construed as limiting the scope of this application. Those skilled in the art can understand that all or part of the processes for implementing the above embodiments and equivalent changes made in accordance with the claims of this application still fall within the scope of this application.

Claims

1. A city path network supplement point optimization method based on an improved Dijkstra algorithm, characterized in that, Includes the following steps: S1: Construct a county-level infrastructure-road network coupled graph model. The node set of the county-level infrastructure-road network coupled graph model includes public service facility nodes, residential community nodes, and road intersection nodes. Mark the real-time load level of each public service facility node and mark the population size weight of each residential community node. S2: The initial reachability and facility load status are calculated using the standard Dijkstra algorithm to obtain the initial coverage and the initial load rate of each facility node; S3: Based on the facility load level and coverage blind spot in the county facility-road network coupling diagram model, construct a load-blind spot coupled gravitational field model, and use the first AI adaptive weight decision model to dynamically adjust the gravitational field weight coefficient to generate candidate supplementary point thermal zones. S4: Introduce coverage gain factor, blind zone coverage reward factor and load balancing factor to construct a multi-dimensional cost function for the improved Dijkstra algorithm, and use the second AI adaptive weight decision model to dynamically adjust the weight coefficients of each factor in the cost function. S5: Traverse each candidate point in the candidate replacement point heat zone, temporarily add the candidate point to the county infrastructure-road network coupling graph model, run the improved Dijkstra algorithm to calculate the comprehensive cost of each candidate point, and select the candidate point with the minimum comprehensive cost as the optimal replacement point; S6: Update the optimal replacement point to the county infrastructure-road network coupled graph model, iteratively execute steps S2 to S5 until the iteration termination condition is met, and output the replacement point optimization scheme. Step S3 specifically includes: The calculation formula for the load-blind zone coupled gravitational field model is as follows: ,in, Points Coverage blind spot density in the area For point The average load level of surrounding facility nodes For point The size of the resident population within the service area, , , These are the gravitational field weighting coefficients; The method of dynamically adjusting the gravitational field weight coefficients using the first AI adaptive weight decision model specifically includes: Input the current coverage, load balancing, population density and scene type characteristics, perform coverage driven correction, load balancing driven correction and population density driven correction, and normalize the corrected weights to meet the conditions ; Step S4 specifically includes: The multidimensional cost function of the improved Dijkstra algorithm is: ,in, Supplementing candidate points The average shortest path time cost to surrounding residential areas and road nodes. For candidate points The increase in coverage after the construction of additional facilities Candidate points The load-blind zone coupling gravity value, For candidate points The extent to which the load on high-load facilities is reduced after the construction of additional facilities. Cost weighting coefficient; The method of dynamically adjusting the weight coefficients of each factor in the cost function using the second AI adaptive weight decision model specifically includes: Input the current coverage gap, load imbalance, population concentration, and scenario type characteristics. Perform coverage-driven correction, load balancing-driven correction, and population density-driven correction, and normalize the corrected weights to meet the conditions. .

2. The urban path network point supplementation optimization method based on the improved Dijkstra algorithm as described in claim 1, characterized in that, Step S1 specifically includes: The real-time load level is the ratio of the resident population covered by the public service facility node to its rated service capacity.

3. The urban path network point supplementation optimization method based on the improved Dijkstra algorithm as described in claim 1, characterized in that, After generating the candidate thermal regions in step S3, the process also includes: A set of candidate replacement points is generated within the candidate replacement point heat zone. The set of candidate replacement points includes locations where facilities can be deployed, such as the middle section of a road, road intersections, and entrances / exits of residential areas.

4. The urban path network point supplementation optimization method based on the improved Dijkstra algorithm as described in claim 1, characterized in that, The iteration termination condition in step S6 is: The county's facility coverage rate has reached a preset threshold, and the facility load balance has reached a preset threshold. Before each iteration, steps S2 to S5 are executed, and the load variation coefficient of each public service facility node is recalculated based on the updated county facility-road network coupling graph model to quantify the facility load balance.

5. The urban path network point supplementation optimization method based on the improved Dijkstra algorithm as described in claim 1, characterized in that, The method further includes: An online interactive visualization platform based on OSM maps is built to display the coverage status, load level, path distribution and coverage blind spots before and after point addition, and generate interactive result files.

6. The urban path network point supplementation optimization method based on the improved Dijkstra algorithm as described in claim 1, characterized in that, After selecting the candidate point with the minimum overall cost as the optimal replacement point in step S5, the following steps are also included: Calculate the overall gain value of the optimal compensation point: ,in, For the increase in coverage, To optimize the load amplitude, , These are the weighting coefficients.

Images