Multi-scenario decision-making method for urban agglomeration ecological resource collaborative configuration network

Abstract

The invention relates to a multi-scenario decision-making method for an urban agglomeration ecological resource collaborative configuration network. The method comprises the following steps: dividing a to-be-researched area into a surplus area and a deficit area according to administrative units; a central point of an administrative unit is used as a network node, a communication edge between adjacent administrative units is constructed, the connection strength of each edge is calculated by considering a corrected gravitation model of a traffic connection level, the connection strength is used as an edge weight, a resource flow potential relationship between cities is comprehensively expressed by the edge weight, and a regional collaborative network is constructed; according to the method, the minimum global network load in different development scenes is taken as a target function, the net traffic Yk of each conduction edge is constrained in each development scene, and meanwhile, the ecological resource demand theoretical value of each development scene on the node level is taken as the target guidance of optimal allocation, a resource allocation model is constructed, and iterative solution is carried out. The method has the adaptive capacity to cope with various strategic development scenes, so that the systematicness and scientificity of ecological resource allocation in the urban agglomeration are improved.

Description

Technical Field

[0001] This invention relates to the field of urban agglomeration ecological resource allocation and regional coordinated development technology, specifically to a multi-scenario decision-making method for an urban agglomeration ecological resource coordinated allocation network, applicable to the optimal allocation of resources between ecological surplus areas and ecological deficit areas within an urban agglomeration. Background Technology

[0002] Against the backdrop of rapid urbanization and coordinated regional development, urban agglomerations have become an important driver of my country's economic growth. However, urban agglomerations generally suffer from uneven development levels and significant differences in resource endowments. The high concentration of population and industry in core cities leads to continuously increasing ecological pressure, while remote, underdeveloped areas possess substantial ecological surpluses. During the development of urban agglomerations, the spatial distance between consumption and supply centers of ecological resources is widening, restricting resource mobility between cities and resulting in low allocation efficiency. Although the overall ecological carrying capacity is declining, a large amount of ecological surplus resources remain underutilized, leading to increasingly prominent resource supply-demand imbalances and significant potential for resource reallocation.

[0003] Regarding inter-city coordination, most related technologies focus on non-physical networks such as policy linkages, horizontal cooperation, or competition mechanisms, emphasizing virtual cooperative relationships between cities. However, the flow of ecological resources depends more on physical connectivity conditions such as transportation infrastructure, as well as the comprehensive economic and social attributes of cities. Currently, in the construction of ecological resource allocation networks in urban agglomerations, integrating the intensity of multi-level physical connections and spatial interactions between cities is a bottleneck, and existing methods struggle to accurately reflect the potential access paths of resources between cities. In terms of modeling ecological resource allocation objectives in urban agglomerations, resource allocation exhibits typical multi-objective conflict characteristics due to multiple factors such as environmental constraints, transportation costs, and resource accessibility. However, existing research often focuses on single objectives and lacks multi-objective trade-off modeling mechanisms under various scenarios. In terms of optimizing ecological resource allocation networks in urban agglomerations, although existing research has employed various intelligent algorithms to solve resource allocation schemes in areas such as carbon emission reduction and ecological restoration, it is still difficult to achieve dynamic collaborative optimization of resource quantity allocation and path selection, and a unified and integrated algorithmic framework is lacking, severely restricting support for regional planning decisions.

[0004] Against this backdrop, the construction of a collaborative network for ecological resources in urban agglomerations faces three technical bottlenecks: First, the spatial network structure of urban agglomerations is complex, requiring unified modeling based on multi-level physical connections and virtual intercity interactions to reflect real resource flow connections; second, there are significant trade-offs among resource allocation objectives, necessitating the establishment of multi-scenario, multi-objective resource allocation models; and third, regional resource allocation scenarios are highly complex, requiring joint optimization of resource quantity and transportation routes under dynamic supply and demand patterns, lacking intelligent resource allocation models that can simultaneously address multiple networks, multiple objectives, and multiple scenarios.

[0005] In summary, how to scientifically integrate the spatial network characteristics of urban agglomerations and construct an optimized decision-making method that takes into account multiple objectives and multi-scenario adaptation has become a key technical challenge for achieving efficient and coordinated allocation of ecological resources in urban agglomerations. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention aims to solve the technical problem of proposing a multi-scenario decision-making method for the coordinated allocation of ecological resources in urban agglomerations. This method can dynamically respond to the ecological resource allocation needs between ecological surplus and ecological deficit areas and has the adaptability to cope with various strategic development scenarios, thereby improving the systematicness and scientific nature of ecological resource allocation within urban agglomerations. The technical solution adopted by this invention to solve the aforementioned technical problem is as follows: A multi-scenario decision-making method for a collaborative allocation network of ecological resources in urban agglomerations includes the following: Obtain the ecological surplus of each administrative unit in the area under study. ER ecological deficit ED The number of administrative units with ecological surplus and ecological deficit was counted separately, and the area to be studied was divided into surplus area and deficit area; Using the central point of an administrative unit as a network node, a Delaunay triangulation is used to construct connecting edges between adjacent administrative units, establishing a spatial topological network that conforms to the principle of geographical proximity. Each edge represents a potential path for the transportation of ecological resources between two administrative units. The connection strength of each edge is calculated by a modified gravity model that considers the level of transportation connections, which serves as the edge weight. The edge weights are used to comprehensively express the potential relationship of resource flow between cities, thus constructing a regional collaborative network. The ecological surplus and ecological deficit of each administrative unit are dynamically considered. The objective function is to minimize the global network load under different development scenarios. Under each development scenario, the net flow of each transmitted "edge" is calculated. Y k A resource allocation model is constructed by imposing constraints and using the theoretical values ​​of ecological resource demand at the "node" level for each development scenario as the target guide for optimal allocation. The resource allocation model is used to optimize the allocation of ecological resources under different development scenarios; The resource allocation model is solved using optimization algorithms to achieve collaborative intelligent decision-making for ecological resources in urban agglomerations under multiple scenarios, namely, to complete the path selection and flow allocation of resources between ecological surplus units and deficit units.

[0007] Furthermore, the optimization algorithm is the NFA-PSO algorithm, which is an improved particle swarm optimization (PSO) algorithm that integrates the network flow allocation mechanism (NFA). It can also be extended to the NFA-GA-HIDMS-PSO algorithm, the NFA-IMOPSO algorithm, or the NFA-HMOPSO algorithm to meet other research needs.

[0008] Furthermore, the NFA-PSO algorithm includes the following steps: In the regional collaborative network, the path selection of ecological resources is the access path for resources to flow from the supply location to the demand location. Each edge represents the resource transportation path between two administrative unit nodes. Based on the edge relationship dataset between the administrative units in space, an adjacency matrix is ​​constructed: in the adjacency matrix, each node represents an administrative unit, and each edge represents the potential path for ecological resource transportation between two administrative units. In the PyCharm platform, the NetworkX graph modeling function is called to transform the adjacency matrix into a graph structure of a regional collaborative network, where nodes correspond to administrative units and edges represent accessible paths between administrative units. Based on the graph structure of the regional collaborative network, the ecological resource surplus value of each node is calculated. ER or deficit value ED Based on this, surplus and deficit units are identified, and a transportation flow matrix is ​​set. F Resource Transportation Ratio Matrix B Under the premise of satisfying the supply and demand balance constraint (i.e., all surplus is allocated to the deficit area), the global network load of the system is minimized. The supply constraint is: the proportion of elements transported in B. b mn The sum does not exceed 1, that is: for each surplus unit m The sum of all its transportation proportions does not exceed 1, a constraint that ensures that the transportation volume of a supply unit cannot exceed its total supply capacity. Demand constraints ensure that the total demand of each demand unit can be met as much as possible; Set the inertia factor, learning factor, and maximum number of iterations for the particle swarm optimization algorithm, and randomly initialize the resource transportation ratio matrix. B ; In each iteration, Dijkstra's algorithm is invoked to dynamically select the shortest path for each pair of supply and demand units. The system prioritizes allocating units with smaller supply and demand ratios that are closest to each other, and updates the resource transportation ratio matrix step by step. B elements in b mn During the step-by-step update process, the indexes of elements that have been allocated are recorded as surplus units. m The set of target deficit units that have been allocated. I mWith deficit units n The set of surplus units that have been allocated J n For transportation ratios that have not yet been updated, i.e., not included I m and J n The elements in the set will form an "unallocated set", which is only valid for... B The unallocated set in the matrix is ​​normalized to ensure that each surplus unit is normalized. m The actual allocation ratio shall not exceed its remaining available capacity, while avoiding exceeding the deficit unit. n The remaining demand; After completing the current step-by-step update allocation, update the matrix. B According to the surplus unit m Allocated to deficit units n Transportation ratio With surplus units m Total supply The product of these components yields the transport flow matrix. F elements in Fl mn Calculate the fitness of the current iteration using the following formula:

[0009] in, Z This represents the overall network load of the system. E Represents the edge relationship dataset of a network graph, | Fl uv | represents the absolute flow on edge (u,v); C uv This represents the connection strength value of the edge (u,v), i.e., the conduction capacity; In each iteration, the fitness values ​​are compared. As the number of iterations increases, the particle positions are continuously adjusted, and the fitness values ​​gradually stabilize. The algorithm eventually converges to the global optimum, achieving coordinated optimization of resource path selection and traffic allocation.

[0010] Furthermore, six development scenarios are defined, and priorities are assigned to each scenario. For development scenario S1, the priority is deficit priority, with resources flowing primarily to regions with severe ecological deficits. For development scenario S2, the priority is economic priority, with resources prioritizing economic development in developed regions. For development scenario S3, the priority is social priority, with resources allocated primarily to densely populated areas. For development scenario S4, the priority is ecological priority, with priority given to regions with high ecological sustainability. For development scenario S5, the priority is vulnerability priority, compensating ecologically vulnerable areas and enhancing overall resilience. For development scenario S6, the priority is comprehensive priority, taking into account ecological, economic, and social factors.

[0011] Furthermore, the net flow of each transmitted "edge" Y n Constraints are imposed, specifically including non-negative lower bound constraints and supply-demand balance constraints; The theoretical values ​​of ecological resource demand at the "node" level under different development scenarios are as follows: For development scenario S1, the theoretical value of ecological resource demand is the total value of all surplus units. m The weighted sum of the ecological resource output capacity, with the weights being the proportion of the deficit in the deficit units; for development scenario S2, all surplus units... m The weighted sum of the ecological resource output capacity, with weights being the regional GDP and deficit amount of the deficit units; for development scenario S3, all surplus units... m The weighted sum of the ecological resource output capacity, with the weights being the total population and deficit of the deficit units; for development scenario S4, all surplus units... m The weighted sum of the ecological resource output capacity, with the weights being the ecological sustainability index and deficit amount of the deficit units; for development scenario S5, all surplus units... m The weighted sum of the ecological resource output capacity, with the weights being the ecological pressure index and deficit amount of the deficit units; for development scenario S6, all surplus units... m The weighted sum of the ecological resource output capacity, with the weights being the ecological pressure index, ecological sustainability index, total population, regional GDP, and deficit amount of the deficit unit.

[0012] Furthermore, the traffic connectivity level is obtained by weighted summation of the traffic network density and traffic accessibility values ​​of each administrative unit within the region. The traffic accessibility value is assigned different values ​​according to different traffic types and different subtypes under each type. Each subtype is divided into four value levels based on distance, and the four value levels are 2, 1.5, 1, and 0 respectively.

[0013] Compared with the prior art, the beneficial effects of the present invention are: The core advantages of this invention include: on the one hand, constructing a regional collaborative network that integrates physical transportation links and non-physical intercity interactions can effectively reflect the potential for resource flow and connectivity between cities, thereby improving the practical applicability and accuracy of the model; on the other hand, this invention is based on multi-scenario decision-making needs and adapts to the resource allocation needs under various development strategies.

[0014] This invention establishes an integrated optimization framework encompassing path selection, flow allocation, and network load. Firstly, the model simultaneously considers the path selection of ecological resources (i.e., the access paths for resources to flow from supply locations to demand locations within the network), resource flow allocation (i.e., resource allocation flow), and the load of network edges (i.e., the flow carried per unit of connectivity for each path). This refers to the resource traffic intensity borne by each path under unit connection strength (i.e., unit transmission capacity), which is used to measure the transmission pressure and efficiency of traffic in the network. It is solved by overall optimization through a unified objective function (i.e., minimizing the global network load), thereby improving resource allocation efficiency and network operation performance.

[0015] This invention proposes the NFA-PSO algorithm, which integrates the Network Flow Allocation (NFA) mechanism and the Particle Swarm Optimization (PSO) algorithm to achieve joint and collaborative optimization of resource allocation quantity and network path selection (i.e., "how much resource to allocate and which path to take"), while minimizing the overall network load. It has the ability, stability and adaptability to solve problems under dynamic changes in resource supply and demand and complex network structures, thus enhancing the model's adaptability and stability under dynamic and complex conditions.

[0016] This invention provides a multi-scenario decision-making approach applicable to the coordinated allocation of ecological resources in urban agglomerations. It integrates regional collaborative network construction, multi-objective modeling, and intelligent optimization solution, and has significant practical application value and methodological innovation. Attached Figure Description

[0017] Figure 1 A comparison chart of the calculated ecological footprint of the Beijing-Tianjin-Hebei urban agglomeration in different years; Figure 2 Calculation results of the ecological carrying capacity of the Beijing-Tianjin-Hebei urban agglomeration; Figure 3 Results of calculation of ecological surplus and deficit in the Beijing-Tianjin-Hebei urban agglomeration; Figure 4 The regional collaborative network structure of the Beijing-Tianjin-Hebei urban agglomeration; Figure 5 Results of the construction of the regional collaborative network of the Beijing-Tianjin-Hebei urban agglomeration; Figure 6 A schematic diagram of the graph structure and adjacency matrix of a regional cooperative network according to one embodiment; Figure 7Spatial configuration results of the ecological resource collaborative network of the Beijing-Tianjin-Hebei urban agglomeration based on the NFA-PSO algorithm. Detailed Implementation

[0018] The present invention will be further explained below with reference to the embodiments and accompanying drawings, but this is not intended to limit the scope of protection of this application.

[0019] The Network Flow Allocation (NFA) mechanism in this invention is a resource allocation mechanism designed based on the characteristics of ecological resource flow, used to improve the Particle Swarm Optimization (PSO) algorithm. The spatial optimization allocation problem of ecological resources is highly complex and dynamic, involving multiple objectives, nonlinearity, and strong constraints; it is a dynamic network flow problem with multiple sources and sinks. The core idea of ​​NFA is based on the supply-demand matching relationship between ecological surplus and deficit areas. By constructing a flow allocation matrix and constraint structure, it expresses the principle of "how to allocate resources and how many resources to allocate to each unit." In solving the multi-objective problem, this invention adopts an improved Particle Swarm Optimization (PSO) algorithm that integrates NFA, namely the NFA-PSO algorithm, effectively improving the solution efficiency and convergence performance of the ecological resource allocation problem. To adapt to different optimization tasks or research needs, this algorithm can be further extended into variants such as the dynamic multi-swarm particle swarm optimization algorithm (GA-HIDMS-PSO) with genetic algorithm-assisted heterogeneous improvement, the improved multi-objective particle swarm optimization algorithm (IMOPSO), and the hybrid multi-objective particle swarm optimization algorithm (HMOPSO), to further improve the performance of the algorithm in multi-objective optimization.

[0020] The resource allocation model of this invention comprises two parts: first, edge selection, which is path selection, i.e., choosing the shortest path; and second, the flow allocation corresponding to each edge. These two parts are co-optimized and cannot be separated. The algorithm updates both flow and path simultaneously in each iteration, demonstrating a "dynamic coordination" effect.

[0021] The present invention provides a multi-scenario decision-making method for urban agglomeration ecological resource collaborative allocation network, comprising the following steps: Step 1: Identification and Accounting of Regional Ecological Gains and Losses 1.1 Ecological Footprint Accounting 1.1.1 Construct an ecological footprint consumption account (Table 1), involving six types of bioproductive land: arable land, forest land, grassland, water area, fossil energy land, and construction land. Consumption types are divided into biological consumption and energy consumption. The classification of consumption items is consistent with the classification of bioproductive land types. Arable land type corresponds to food crops, beans, tubers, hemp, sugar beets, vegetables, fungi, pork, poultry meat, and eggs, etc.; forest land type corresponds to fruits and timber, etc.; grassland type corresponds to beef and mutton, etc.; water area type corresponds to aquatic products; fossil energy land type corresponds to coal, oil, natural gas, gasoline, and diesel consumption items; and construction land type corresponds to electricity consumption items.

[0022] Table 1 Ecological Footprint Consumption Account

[0023] 1.1.2 The ecological footprint of each administrative unit in the urban agglomeration is calculated using the following formula:

[0024] In the formula, EF Ecological footprint (ha). N This represents the total population (in people) of a certain administrative unit. ef This represents the average ecological footprint per capita (ha / cap). i The type of consumption item; j Types of biologically productive land; aa i for i Bioproductive area (hectares, ha) converted from consumer items; r j As a balancing factor for various types of land; c i yes i Per capita consumption of this type of consumer item (kg / cap); p i for i National average annual output of consumer items (kg·ha) -1 ·a -1 The calculation results are shown below. Figure 1 .

[0025] 1.2 Ecological Carrying Capacity Accounting The area of ​​biologically productive land within the accounting unit that sustains human survival and development. The calculation formula is as follows:

[0026] In the formula, ECC Ecological carrying capacity (ha). ecc Ecological carrying capacity per capita (ha / cap). j As a type of biologically productive land,N This represents the total population (in people) of a certain administrative unit. a j for j Area of ​​biodiversity-producing land (ha / cap). r j and y j These are the equilibrium factor and the yield factor, respectively. The calculation results are shown below. Figure 2 .

[0027] 1.3 Accounting for Ecological Surplus and Deficit Ecological surplus and ecological deficit are determined by ecological carrying capacity ( ECC ) and ecological footprint ( EF The difference is calculated using the following formula:

[0028] In the formula, u Indicates a certain administrative unit, when When, it indicates an ecological surplus ( ER );when When, it indicates an ecological deficit ( ED ); ECC u Indicates administrative unit u Ecological carrying capacity; EF u Indicates administrative unit u Ecological footprint. Calculation results are shown below. Figure 3 .

[0029] Step 2: Construction of Regional Collaborative Network 2.1 Network Structure Modeling 2.1.1 Establishing Network Nodes and Edges: Using the central point of an administrative unit as a network node, a Delaunay triangulation is used to construct connecting edges between adjacent administrative units, establishing a spatial topological network that conforms to the principle of geographical proximity. In an empirical case study of the Beijing-Tianjin-Hebei urban agglomeration, a regional collaborative network structure covering 199 nodes and 582 edges was constructed, such as... Figure 4 As shown.

[0030] 2.1.2 Setting Edge Weights: The connection strength of each edge is calculated using an improved modified gravity model. C uv As edge weights, this process introduces multi-layered "physical" transportation connections, assigning values ​​according to multi-level transportation infrastructure. It also considers and integrates physical transportation connections with non-physical (GDP and population) intercity interactions, and incorporates the transportation connection levels of each administrative unit and the regional average transportation connection level to comprehensively measure the connection strength of each edge. The connection strength of each edge... Cuv The value of the edge weight determines the transmission capacity of the network path; the larger the edge weight, the higher the ability to transfer ecological resources. The formula is as follows:

[0031] In the formula, C uv For administrative units u and administrative units v The strength of the connection between them T u For administrative units u The level of transportation connectivity, This represents the average level of regional transportation connectivity. P u Population (people) of administrative unit u. G u For administrative units u GDP (in ten thousand yuan) L uv The distance between the two administrative units is expressed in meters (Euclidean distance).

[0032] Specifically, the "level of traffic connectivity" is calculated by weighted summation of two indicators: road network density and accessibility (Table 2) for each administrative unit within the region. The formula is as follows:

[0033] In the formula: T u Indicates administrative unit u The level of transportation connectivity; D u For administrative units u The traffic network density value; H u For administrative units u Accessibility of transportation; w 1 , w 2 These represent the weights corresponding to the two indicators, respectively. This embodiment uses a proportional weighting method, therefore taking... .

[0034] This invention constructs a multi-layered physical connection, namely a spatial connection network between cities built by different levels of transportation infrastructure (such as high-speed railways, intercity railways, ordinary roads, and expressways). The physical connections between cities are divided into multiple layers (see Table 2, including expressways, national highways, provincial highways, etc.), each layer representing different intensities, speeds, or levels of traffic accessibility. The connection strength between cities is calculated through a weighting method, which facilitates the construction of a graph structure for a regional collaborative network in subsequent algorithms, used to simulate the potential flow channels of ecological resources under different physical path conditions.

[0035] The "Standards" and "Assignments" in the third and fourth columns of Table 2 are an improved indicator system based on the spatial scope and actual distances of urban agglomerations. Specifically, the grading standards for "Distance" in the third column are adjusted according to the spatial scale, facility layout, and accessibility characteristics of the urban agglomeration. Furthermore, the "Assignments" in the fourth column are adjusted and weighted based on existing empirical assignments, taking into account the impact of different transportation facilities within the region on resource flow capacity. For example, the accessibility weight of highways and railway stations is the highest (assigned a value of 2), reflecting their dominant role in resource allocation; while the weights of secondary roads and long-distance facilities decrease sequentially, reflecting the diminishing marginal impact on flow potential.

[0036] Table 2. Traffic Accessibility Evaluation Indicators

[0037] 2.1.3 Taking the Beijing-Tianjin-Hebei urban agglomeration as an empirical case, the resulting regional collaborative network is as follows: Figure 5 As shown, it covers 199 nodes, 582 edges, and their corresponding edge weights.

[0038] This invention constructs a regional collaborative network that uses edge weights to comprehensively express the potential relationships of resource flows between cities. The edge weights are calculated using a modified gravity model and integrate two key elements: first, the strength of physical connections based on transportation links, reflecting the spatial accessibility between cities; and second, the intensity of virtual intercity interaction, measured by socio-economic factors such as resident population, GDP, and resource endowment, reflecting the attractiveness of resource supply and demand. This network can effectively characterize the potential input-output relationships and collaborative strength of ecological resources between cities.

[0039] Step 3: Setting up multiple scenarios and constructing the objective function 3.1 Define multi-scenario decision-making objectives 3.1.1 Six development scenarios were defined (Table 3) to simulate various possible paths for future urban development, and priorities were assigned to each scenario. These scenarios reflect the coordination and trade-offs among ecological, economic, social, and other factors in different cities.

[0040] Table 3. Description of Optimization Objectives for Multiple Scenarios

[0041] 3.2 Constructing a resource allocation model 3.2.1 Establish the objective function. The construction of the objective function should not only dynamically consider the ecological surplus and ecological deficit of each administrative unit, but also minimize the global network load according to different scenarios. Z At the same time, it strives to meet theoretical values ​​and supply-demand balance as much as possible, thereby achieving efficient allocation of ecological resources. The objective function formula is as follows:

[0042] In the formula, For the first k The processing of the absolute value of the cumulative flow of each side can effectively avoid the positive and negative cancellation errors caused by the difference in the direction of resource flow, and reflect the true flow carrying capacity of each side. C k For the first k The connection strength value of each edge represents the conductivity. k This represents the number of edges in the regional collaborative network.

[0043] 3.2.2 Set constraints and, under each development scenario, determine the net flow of each transmitted "edge". The constraints are applied, as shown in the following formula: (1) Non-negative lower bound constraint:

[0044] (2) Supply and demand balance constraints:

[0045] in, S m Represents surplus units m Ecological resource output capacity (unit, ha). D n For deficit units n The ecological resource input capacity (unit: ha). That is, the surplus in the surplus area flows entirely into the deficit area.

[0046] Calculate the theoretical values ​​of ecological resource demand at the "node level" for each development scenario (deficit unit). n The allocated input flow serves as the target guide for model optimization allocation, as shown in Table 4. Under the dual constraints of ecological resource constraints and network transmission complexity (in the application of Network Flow Allocation (NFA) mechanisms, resource flow allocation not only depends on the resource supply and demand relationship between nodes, but also on the mutual relationships between nodes in the spatial network, the mutual influence of resource flow paths, and the overall network load), the model ultimately solves for the cumulative flow at the edge level.Fl This will cause the input flow of the red byte point to gradually converge to the optimal approximation solution of its theoretical requirement value.

[0047] Table 4 Theoretical values ​​of ecological resource demand at each development scenario node level

[0048] By setting objective functions and constraints, the system considers multiple factors such as ecological pressure spillover, resource replenishment, and transportation costs. The resource allocation model can achieve optimal allocation of ecological resources under different development scenarios. The resource allocation model in this invention fully considers the complex trade-offs between multiple objectives, incorporating "path selection—flow configuration—network load" into a unified optimization framework. Six development scenarios (S1–S6) consider potential trade-offs between external ecological, economic, and social factors, simulating various possible strategies for future urban development, thus enhancing the model's adaptability and decision-making applicability in complex situations.

[0049] Step 4: Optimization and solution of resource allocation model based on NFA-PSO 4.1 Data Modeling 4.1.1 Establishing a Edge Relationship Dataset. Based on the regional collaborative network constructed in the GIS platform (Step 2), the nodes of each spatial administrative unit are numbered, and the connection edge information of each spatial administrative unit is extracted to form a edge relationship dataset. Each edge represents a resource transportation path between two cities, and the edge naming format is "Str (starting point)_End (ending point)". Taking the Beijing-Tianjin-Hebei urban agglomeration as an empirical case (Table 5), for example, "138_148" represents "Anci District – Wuqing District".

[0050] Table 5. Partial dataset of edge relationships of spatial units in the Beijing-Tianjin-Hebei urban agglomeration.

[0051] 4.1.2 Constructing the Graph Structure of the Regional Collaborative Network. Based on the edge relationship dataset between administrative units, an adjacency matrix is ​​constructed. In the adjacency matrix, each node represents an administrative unit, and each edge represents a potential path for ecological resource transportation between two administrative units. Subsequently, the NetworkX graph modeling function is called in the PyCharm platform to transform the adjacency matrix into the graph structure of the regional collaborative network, where nodes correspond to administrative units, and edges represent accessible paths between administrative units (see...). Figure 6 (in the figure) N n Represents a cell node. e ij Indicates connection unit i and unit j The edges in the matrix bij Represents the weight value of the edge. ∞ This indicates that there is no connection between the two nodes.

[0052] 4.2 Solution Process Integrating Particle Swarm Optimization (PSO) and Network Flow Assignment (NFA) algorithms, and jointly using Dijkstra's algorithm, an NFA-PSO optimization model is constructed. This NFA-PSO optimization model addresses the resource transport ratio matrix. B Iterative optimization is performed to minimize the global network load of the system, completing resource allocation and path selection between ecological surplus units and deficit units. The optimization objective of the NFA-PSO algorithm can be expressed as: achieving the optimal resource transportation ratio matrix. B Under the premise of satisfying the supply and demand balance constraint (i.e., all surplus is allocated to the deficit area), the global network load of the system is minimized.

[0053] The specific process of the NFA-PSO optimization model is as follows: 4.2.1 Multi-objective problem modeling and variable setting: Based on the graph structure of the regional collaborative network, the ecological resource surplus value of each node is calculated ( ER ) or deficit value ( ED This allows for the identification of surplus and deficit units. In surplus units... ER >0 indicates the amount of resources that can be allocated in this unit; in deficit units... ED A value less than 0 indicates the quantity of resources required by that unit. A transportation flow matrix is ​​defined based on the number of surplus and deficit units. F Resource Transportation Ratio Matrix B This decouples resource allocation tasks from the dimension of [the resource allocation task].

[0054] 4.2.2 Constraint Transformation: Transportation Flow Matrix F Resource Transportation Ratio Matrix B The dimension is ,in x It is the number of surplus units. y It is the number of deficit units. B It is the core optimization variable in the Particle Swarm Optimization (PSO) algorithm, representing a candidate solution for a resource transportation allocation scheme. B It is a dynamic optimization parameter matrix whose element values ​​are continuously updated in the optimization iteration to approximate the optimal solution, which represents the actual transportation ratio of each spatial administrative unit.

[0055] Set parameters such as inertia factor, learning factor, and maximum number of iterations for the particle swarm optimization algorithm, and randomly initialize the resource transportation ratio matrix. B ; The constraints of the particle swarm optimization algorithm are determined in conjunction with the network flow allocation mechanism (NFA), specifically: (1) Supply constraints, in the resource transportation ratio matrix B, the elements b mn Indicates from surplus unit m Towards the deficit unit n The transportation ratio (i.e., the element in the m-th row and n-th column of matrix B). b mn The sum of the total does not exceed 1, thereby ensuring that the maximum transportation capacity of each unit is not exceeded, while maintaining the relative relationship of the original allocation trend; (2) Demand constraint, conduct demand constraint verification, and ensure that the total demand of the demand unit (deficit unit) can be met as much as possible.

[0056] (1) Supply constraints:

[0057] In the formula, b mn To extract from surplus units m Towards the deficit unit n The proportion of transportation, y The number of deficit units, x This represents the number of surplus units.

[0058] For each surplus unit m The total proportion of all its transportation does not exceed 1, a constraint that ensures that the transportation volume of a supply unit cannot exceed its total supply capacity.

[0059] (1) Demand constraints:

[0060] In the formula, S m Indicates the first m The total supply of each surplus unit (i.e., the current surplus resources). D n Representing the n The total demand of each deficit unit (i.e., the current surplus demand). x The number of surplus units; y This represents the number of deficit units.

[0061] This constraint ensures that the total demand of the demand unit can be met as much as possible.

[0062] In each iteration, Dijkstra's algorithm is invoked to dynamically select the shortest path for each pair of supply and demand units. The system prioritizes allocating units with smaller supply and demand ratios that are closest to each other, and updates the resource transportation ratio matrix step by step. B The elements in the set. During the step-by-step update process, the indices of the elements that have been allocated are recorded as a set. Im (Indicates surplus unit) m The set of target deficit units that have been allocated, namely: m Which ones have already been given? n (Resources allocated) and J n (Indicates a deficit unit) n The set of surplus units that have been allocated, namely: n From which m (Resources received). For shipping ratios that have not yet been updated (i.e. not included). I m and J n The elements in the set will form an "unassigned set", which is only applicable to... B The unallocated set in the matrix is ​​normalized to ensure that each surplus unit is normalized. m The allocation ratio shall not exceed its remaining supply capacity (i.e., the remaining allocation ratio of surplus units, which is the formula below). r m ), while avoiding exceeding the deficit unit n The remaining demand (i.e., the percentage of remaining demand in the deficit unit, which is the formula below) r n After completing the current step-by-step update allocation, update the matrix. B The population for the next iteration is obtained by normalization.

[0063] The formulas for calculating the remaining allocable proportion and the remaining demand proportion are as follows:

[0064]

[0065] In the formula, r m For surplus units m The remaining distributable proportion, r n For deficit units n The percentage of remaining demand, I m and J n This represents the set of indices of the elements that have already been filled. b mn Indicates from surplus unit m Allocated to deficit units n The proportion of transportation; S m For surplus units m Total supply (ha). D n For deficit unitsn Total demand (ha).

[0066] In update B Then, the transport flow matrix is ​​obtained according to the following formula. F elements in Fl mn Used for subsequent fitness calculations:

[0067] in Fl mn Indicates from surplus unit m Towards the deficit unit n Actual cumulative transport volume, transport volume matrix F It directly reflects the actual flow of resources between various spatial units.

[0068] 4.2.3 Iterative Update and Fitness Calculation: In each iteration, after the particle updates its position, it must satisfy the supply and demand constraints to ensure that its allocation scheme (resource transportation ratio matrix B) is effective.

[0069] To ensure the correctness of the flow direction, the cumulative flow must take direction into account. If the flow direction is consistent with the defined direction of the edge, it is a positive value; if the direction is opposite, it is a negative value. In this way, the cumulative flow of each edge in the network can fully reflect the resource transportation flow between supply and demand.

[0070] The fitness calculation formula is the system's global network load Z:

[0071] In the formula, E Represents the edge relationship dataset of a network graph, | Fl uv | represents the absolute flow on edge (u,v), where the absolute value can prevent the erroneous accumulation of reverse flow in the network; C uv This represents the connection strength value of the edge (u,v), i.e., the conduction capacity.

[0072] In each iteration, the fitness is compared to see if it is minimized. As the number of iterations increases, the particle position is continuously adjusted, the fitness function gradually converges, and finally the global optimal solution is obtained, realizing the collaborative optimization of resource path selection and traffic allocation.

[0073] The NFA-PSO optimization model in this invention has the following advantages: (1) Dynamically optimized resource transportation ratio matrix B . It serves as a dynamically optimized parameter matrix in the model, rather than simply a static setting. The model is dynamically adjusted as the optimization process progresses, rather than being pre-fixed. This allows the model to adaptively adjust to changes in demand, supply, and network structure in real-world applications, thereby achieving a more suitable resource allocation.

[0074] (2) Constraints and normalization. The created constraints are met, and the rationality of resource flow is ensured through normalization and constraints.

[0075] (3) A multi-objective "decomposition-cooperation" strategy is adopted. The optimization problem is decomposed into multiple cooperative single-objective optimization sub-tasks, covering the flow allocation of the network graph and the dynamic cooperative coupling of Dijkstra's algorithm and PSO algorithm. This is an innovative optimization of the existing transportation flow allocation method. Among them, PSO iteratively optimizes the flow allocation ratio of "supply node" to "edge", and Dijkstra's algorithm dynamically updates the shortest path and then calculates the network load. Through the interaction and coordination of sub-tasks, the supply and demand balance and the minimization of the global network load are finally achieved.

[0076] 4.2.4 Taking the Beijing-Tianjin-Hebei urban agglomeration as an empirical case, the maximum number of iterations was 100, the inertia factor was set to 0.9, and both the individual experience factor and the group experience factor were set to 1.5. In each iteration, Dijkstra's algorithm was called to update the transportation path between each pair of supply and demand cities, and the resource flow direction was determined according to the shortest path distance to complete the network load calculation and resource scheduling.

[0077] The final output is the allocation scheme (i.e., the minimum cost scheme) corresponding to the minimum global network load obtained after multiple rounds of iteration and convergence. The spatial resource allocation results of the ecological resource coordination network of the Beijing-Tianjin-Hebei urban agglomeration in 2020 are as follows: Figure 7 As shown.

[0078] Any aspects not covered in this invention are applicable to existing technologies.

Claims

1. A multi-scenario decision-making method for a coordinated configuration network of ecological resources in urban agglomerations, characterized in that, The method comprises the following contents: Obtaining ecological surplus of each administrative unit in the region to be studied ER and ecological deficit ED , respectively, the number of surplus and deficit administrative units is counted, and the region to be studied is divided into surplus and deficit areas; The center point of the administrative unit is taken as a network node, a Delaunay triangular network is used to construct a connected edge between adjacent administrative units, a spatial topological network conforming to the principle of geographical adjacency is established, each edge represents a potential path of ecological resource transportation between two administrative units, a modified gravity model considering the traffic connection level is used to calculate the connection strength of each edge as an edge weight, and the edge weight is used to comprehensively express the potential resource flow relationship between cities, and a regional collaborative network is constructed; The ecological surplus and the ecological deficit of each administrative unit are dynamically considered, the global network load is minimized as the objective function under different development scenarios, and the net flow of each transmission "edge" is calculated under each development scenario Y k The resource allocation model is constructed by performing constraints and using the theoretical ecological resource demand value of each development scenario at the "node" level as the target guide for optimal allocation. The resource allocation model is used to realize the optimal allocation of ecological resources under different development scenarios; An optimization algorithm is used to solve the resource allocation model, and the collaborative intelligent decision of the ecological resources of the urban agglomeration under multiple scenarios is realized, that is, the path selection and flow distribution of resources between the ecological surplus units and the deficit units are completed.

2. The method of claim 1, wherein, The optimization algorithm is an NFA-PSO algorithm, an NFA-GA-HIDMS-PSO algorithm, an NFA-IMOPSO algorithm or an NFA-HMOPSO algorithm to adapt to other research needs.

3. The method of claim 2, wherein, The NFA-PSO algorithm comprises the following steps: In the regional collaborative network, the path selection of the ecological resources is the accessible path of the resources flowing from the supply place to the demand place, each edge represents a resource transportation path between two administrative unit nodes, an adjacency matrix is constructed according to the edge relationship data set between the administrative units in space, in the adjacency matrix, each node represents an administrative unit, and each edge represents a potential path of ecological resource transportation between two administrative units; In the PyCharm platform, a NetworkX graph modeling function is called to convert the adjacency matrix into a graph structure of the regional collaborative network, wherein the nodes correspond to the administrative units, and the edges represent the accessible paths between the administrative units. Based on the graph structure of the regional collaborative network, the ecological resource surplus value of each node is calculated. ER or deficit value ED Based on this, surplus and deficit units are identified, and a transportation flow matrix is ​​set. F Resource Transportation Ratio Matrix B Under the premise of satisfying the supply and demand balance constraint (i.e., all surplus is allocated to the deficit area), the global network load of the system is minimized. The supply constraint is: the proportion of elements transported in B. b mn The sum does not exceed 1, that is: for each surplus unit m The sum of all its transportation proportions does not exceed 1, a constraint that ensures that the transportation volume of a supply unit cannot exceed its total supply capacity. The demand constraint ensures that the total demand of the demand unit can be satisfied as much as possible; The inertia factor, learning factor and maximum iteration number of the particle swarm algorithm are set, and the resource transportation proportion matrix is randomly initialized B ; In each iteration, Dijkstra algorithm is called to dynamically select the shortest path for each pair of supply and demand units, and the system preferentially allocates the units with smaller supply and demand and the nearest units, and step-by-step updates the resource transportation proportion matrix B in each iteration b mn In the step-by-step updating process, the element index of the completed allocation is recorded as a surplus unit m a set of target deficit units of the completed allocation I m a set of deficit units n a set of surplus units that have received allocation J n ; for the transportation proportion that has not been updated, i.e. not contained in I m and J n , a "non-allocated set" is formed, and normalization is only performed on B the non-allocated set in the matrix to ensure that the actual allocation proportion of each surplus unit m does not exceed its remaining supply capacity, while avoiding exceeding the remaining demand of the deficit unit n ; After the current sub-step update allocation is completed, the update matrix B is obtained by multiplying the transport ratio m from the surplus unit n to the deficit unit by the total supply amount m of the surplus unit to obtain the element F in the transport flow matrix Fl mn The fitness of the current iteration is calculated according to the following formula: ， wherein, Z is the system global network load; E denotes the edge relation data set of the network graph, Fl uv denotes the absolute flow on edge (u, v); C uv denotes the contact strength value, i.e. the conductance, of edge (u, v); The size of the fitness is compared at each iteration, with the particle position being adjusted and the fitness value being gradually stabilized with the increase of the iteration number, so that the algorithm converges to the global optimal solution, and the collaborative optimization of the path selection and flow distribution of the resources is realized.

4. The method of claim 1, wherein, Six types of development scenarios are set, and priority settings are made for each scenario, for the development scenario S1, the priority setting is deficit priority, and the resources preferentially flow to the areas with serious ecological deficit, for the development scenario S2, the priority setting is economic priority, and the resources preferentially guarantee the economic development of developed areas, for the development scenario S3, the priority setting is social priority, and the resources are preferentially allocated to densely populated areas, for the development scenario S4, the priority setting is ecological priority, and the resources are preferentially supported to areas with high ecological sustainability, for the development scenario S5, the priority setting is vulnerability priority, and the ecological vulnerability areas are compensated to improve the overall resilience, and for the development scenario S6, the priority setting is comprehensive priority, and the ecological, economic and social factors are comprehensively considered.

5. The method of claim 1, wherein, Net flow for each transmission "edge" Y n Constraints are imposed, including non-negative lower bound constraints and supply-demand balance constraints. The theoretical values ​​of ecological resource demand at the "node" level under different development scenarios are as follows: For development scenario S1, the theoretical value of ecological resource demand is the total value of all surplus units. m The weighted sum of the ecological resource output capacity, with the weights being the proportion of the deficit in the deficit units; for development scenario S2, all surplus units... m The weighted sum of the ecological resource output capacity, with weights being the regional GDP and deficit amount of the deficit units; for development scenario S3, all surplus units... m The weighted sum of the ecological resource output capacity, with the weights being the total population and deficit of the deficit units; for development scenario S4, all surplus units... m The weighted sum of the ecological resource output capacity, with the weights being the ecological sustainability index and deficit amount of the deficit units; for development scenario S5, all surplus units... m The weighted sum of the ecological resource output capacity, with the weights being the ecological pressure index and deficit amount of the deficit units; for development scenario S6, all surplus units... m The weighted sum of the ecological resource output capacity, with the weights being the ecological pressure index, ecological sustainability index, total population, regional GDP, and deficit amount of the deficit unit.

6. The method of claim 1, wherein, The traffic contact level is obtained by weighted summation of the traffic network density and the traffic accessibility of each administrative unit in the region, wherein the value of the traffic accessibility is respectively given different values according to different traffic types and different subtypes under each type, and each subtype is divided into four value grades according to the distance, and the four value grades are 2, 1.5, 1 and 0 in turn.

Images