A multi-objective collaborative optimization design method for sponge parks based on LID technology

Abstract

The present invention relates to the technical field of computer-aided design and intelligent optimization, and in particular to a multi-objective collaborative optimization design method for sponge parks based on LID technology. A parametric database of LID facilities is constructed by collecting multi-source data such as geospatial and hydrometeorological data; a multi-objective collaborative optimization model covering hydrological, ecological, economic, and social benefits is established, and solved using a hybrid intelligent optimization algorithm composed of an improved NSGA-II algorithm and an enhanced particle swarm optimization algorithm to generate a Pareto optimal solution set; the SWMM and GIS coupling technology and the improved TOPSIS method are used to screen the comprehensive optimal solution; an Internet of Things monitoring network and a digital twin system are deployed, and according to trigger conditions such as runoff coefficient deviation and vegetation coverage change, the adaptive dynamic optimization of the design scheme is realized. The present invention breaks through the limitations of traditional single objectives, and significantly improves the comprehensive benefits of sponge parks and ensures their long-term stable operation in complex environments through multi-objective collaboration, intelligent algorithm optimization, and dynamic adjustment.

Description

Technical Field

[0001] The present invention relates to the technical field of computer-aided design and intelligent optimization, and specifically provides a multi-objective collaborative optimization design method for sponge parks based on LID technology. Background Technique

[0002] To address urban water problems, the concept of sponge cities has emerged. Sponge cities emphasize combining natural and artificial means to construct a series of facilities with functions of water absorption, storage, infiltration, and purification, such as permeable pavements, rain gardens, and ecological wetlands, enabling the city to be like a "sponge" and having good "resilience" in adapting to environmental changes and coping with natural disasters. As an important part of sponge city construction, sponge parks can not only provide green spaces for urban residents to relax and entertain but also play an important role in rainwater storage and ecological purification, becoming one of the effective ways to solve urban water problems.

[0003] However, there are still some problems in the current design and construction of sponge parks. Traditional sponge park designs often focus on single objectives, such as simple rainwater collection or landscape creation, lacking comprehensive consideration of multiple benefits in aspects such as hydrology, ecology, economy, and society. In terms of design methods, although some numerical simulation and evaluation technologies have been applied, most are not accurate and comprehensive enough to accurately quantify the comprehensive benefits of different design schemes. At the same time, due to the lack of real-time monitoring and dynamic optimization mechanisms, sponge parks are difficult to adjust and optimize in a timely manner during actual operation in the face of complex and changing environmental conditions and urban development needs, resulting in the inability to fully utilize their functions and achieve long-term stable multi-objective collaborative optimization.

[0004] Therefore, a multi-objective collaborative optimization design method for sponge parks based on LID technology is proposed to address the above problems. Summary of the Invention

[0005] The purpose of the present invention is to provide a multi-objective collaborative optimization design method for sponge parks based on LID technology to solve the problems raised in the above background technique.

[0006] To achieve the above purpose, the present invention provides the following technical solutions:

[0007] A multi-objective collaborative optimization design method for sponge parks based on LID technology includes the following steps:

[0008] S1. Data collection and processing: Obtain the geospatial data, hydrometeorological data, ecological environment data, and social economic data of the target area, and construct a parametric database of LID facilities;

[0009] S2. Model construction: Establish a multi-objective collaborative optimization model including hydrological benefits, ecological benefits, economic benefits, and social benefits;

[0010] S3. Algorithm Solving: The multi-objective collaborative optimization model is solved using a hybrid intelligent optimization algorithm to generate a Pareto optimal solution set;

[0011] S4. Solution Verification: The comprehensive optimal solution is screened through coupled numerical simulation and a multi-dimensional evaluation system;

[0012] S5. Dynamic Optimization: An Internet of Things monitoring network is deployed and a digital twin system is established to achieve adaptive dynamic optimization of the design solution.

[0013] As a preferred solution, in step S1:

[0014] The geospatial data includes a digital elevation model with a resolution ≤ 0.5 m, a spatial distribution map of soil permeability coefficient with a sampling interval ≤ 10 m, and a rainfall intensity matrix of historical rainstorm events;

[0015] The LID facility parameterization database contains a dynamic correction coefficient for the effective porosity of permeable pavement. The dynamic correction coefficient is adjusted annually according to the number of operating days by the β coefficient, where the value range of β is from 0.15 to 0.25.

[0016] As a preferred solution, in step S2, the multi-objective collaborative optimization model in step S2 satisfies:

[0017] The hydrological benefit objective is quantified by the weighted sum of the surface runoff reduction rate, rainwater retention time, and peak flow delay coefficient. The weight coefficients α1, α2, α3 satisfy α1 + α2 + α3 = 1;

[0018] The ecological benefit objective is quantified by the weighted sum of the vegetation coverage rate, biodiversity index, and carbon sequestration capacity. The weight coefficients β1, β2, β3 satisfy β1 + β2 + β3 = 1.

[0019] As a preferred solution, the hybrid intelligent optimization algorithm in step S3 includes:

[0020] The improved NSGA-II algorithm, and its improvements include:

[0021] The crossover probability is dynamically adjusted. The crossover probability value changes periodically according to the cosine function with the number of iterations, and the change range is from 0.5 to 0.9;

[0022] The elite retention strategy based on crowding entropy, and the top 15% of non-dominated solutions with crowding entropy values are retained in each generation;

[0023] The enhanced particle swarm optimization algorithm, and its velocity update includes a non-linear decreasing inertia weight term. The inertia weight decays from the initial value of 0.9 to 0.4 according to the exponential function.

[0024] As an optimal solution, the multi-objective collaborative mechanism of the hybrid intelligent optimization algorithm includes:

[0025] A dynamic weight allocation function, where the weights of each objective are dynamically adjusted according to the exponential function of the normalized fitness value, and the adjustment coefficient η ranges from 0.8 to 1.2;

[0026] The adaptive penalty function method, where the penalty factor increases exponentially with the number of iterations, and the initial penalty factor ranges from 10 3 to 10 4 .

[0027] As an optimal solution, in step S4:

[0028] The coupled numerical simulation uses the SWMM model and GIS spatial analysis technology to calculate the runoff volume through coupled calculation, and the calculation process includes the integral operation of the dynamic infiltration rate function and the roughness coefficient correction coefficient;

[0029] The multi-dimensional evaluation system uses the improved TOPSIS method. The discrimination coefficient ρ is introduced in the calculation of the closeness degree, and its value range is from 0.4 to 0.6, and the closeness to the ideal solution is measured based on the Mahalanobis distance metric scheme.

[0030] As an optimal solution, the Internet of Things monitoring network in step S5 includes:

[0031] A distributed hydrological sensor array, including a laser raindrop spectrometer with an accuracy of ±5%, a capacitive soil moisture sensor with a range of 0 - 100%, and an ultrasonic water level gauge with a resolution of 1 mm;

[0032] An edge computing node, which builds a long short-term memory model to calculate the runoff coefficient deviation value in real time , The calculation formula is the absolute percentage deviation between the measured value and the designed value;

[0033] A data transmission module, which supports the LoRaWAN protocol and has a packet loss rate ≤ 0.1%.

[0034] As an optimal solution, the triggering conditions for adaptive dynamic optimization are:

[0035] The runoff coefficient deviation value exceeds 15% for 3 consecutive sampling periods;

[0036] The vegetation coverage rate drops by more than 20% compared with the designed value;

[0037] The rainstorm recurrence period exceeds 1.3 times the design standard;

[0038] The optimization response time meets the requirements of ≤ 48 hours in the normal mode and ≤ 6 hours in the emergency mode.

[0039] As a preferred solution, the hybrid intelligent optimization algorithm introduces a multi-modal strategy switching mechanism:

[0040] When the population diversity index H > 0.7, enable the global search mode and set the particle swarm inertia weight > 0.8;

[0041] When the convergence speed < 0.05, enable the local enhancement mode and set the mutation probability > 0.15;

[0042] When the Pareto front change rate δ < 0.01, trigger chaotic perturbation, and the perturbation formula is , where, is the new position generated after chaotic perturbation, is the original position, is the chaos parameter, takes values from 3.8 to 4.0.

[0043] As a preferred solution, the digital twin system includes:

[0044] The physical entity layer deploys LID facility status sensors with a density ≥ 3 units / 100 m²;

[0045] The virtual model layer integrates a SWMM-HEC coupled simulation engine with a time step ≤ 1 minute;

[0046] The data interaction layer uses the OPC UA protocol to achieve real-time synchronization with a delay ≤ 200 ms;

[0047] The decision optimization layer embeds a mixed integer programming algorithm for full life cycle cost optimization.

[0048] It can be seen from the technical solutions provided by the present invention above that for a multi-objective collaborative optimization design method of a sponge park based on LID technology provided by the present invention, the beneficial effects are:

[0049] Multi-objective collaborative optimization to improve comprehensive benefits: By constructing a multi-objective collaborative optimization model covering hydrological benefits, ecological benefits, economic benefits and social benefits, the limitation of traditional single-objective design is broken; in terms of hydrology, indicators such as runoff reduction rate and rainwater retention time are accurately quantified, effectively alleviating urban waterlogging; ecologically, guided by vegetation coverage rate, biodiversity, etc., the stability of the ecosystem is enhanced; both economic and social benefits take into account construction and operation costs and public needs, achieving the maximization of the comprehensive benefits of the sponge park;

[0050] Intelligent algorithm innovation to improve optimization efficiency and quality: A hybrid intelligent optimization algorithm composed of an improved NSGA-II algorithm and an enhanced particle swarm optimization algorithm is adopted, combined with a dynamic weight allocation, an adaptive penalty function, and a multi-modal optimization strategy switching mechanism. During the solution process, it can both globally search and explore a vast solution space and locally optimize to improve the accuracy of the solution; compared with traditional algorithms, it significantly improves the solution efficiency, avoids falling into local optima, and the generated Pareto optimal solution set provides richer and higher-quality choices for the design solution;

[0051] Data-driven and dynamic optimization to enhance adaptability: Rely on the Internet of Things monitoring network to collect multi-source data such as geospatial and hydro-meteorological in real time, and construct a parameterized database of LID facilities; combined with the digital twin system, realize real-time monitoring and accurate simulation of the operation status of the sponge park; when situations such as too large a deviation in runoff coefficient, a vegetation coverage rate drop exceeding the threshold, and a storm return period exceeding the standard occur, automatically trigger adaptive dynamic optimization, with hierarchical responses for regular optimization and emergency optimization, ensuring that the park always maintains a good operation status under different environmental conditions, and significantly improving the adaptability and reliability of the system;

[0052] Precise evaluation and scientific decision-making to ensure the feasibility of the solution: Through coupling the SWMM model and GIS spatial analysis technology for numerical simulation, combined with the improved TOPSIS method to construct a multi-dimensional evaluation system, quantitatively evaluate and screen the design solution from multiple aspects; avoid subjective judgment, ensure that the selected comprehensive optimal solution has solid data support and scientific basis, and has higher feasibility and effectiveness in actual engineering applications;

[0053] Full life cycle management to promote sustainable development: The full life cycle digital twin system realizes the full-process digital management of the sponge park from construction to operation and maintenance; the real-time interaction between the physical entity layer and the virtual model layer, combined with the intelligent algorithm of the decision optimization layer, provides guarantee for the long-term stable operation of the park, reduces resource waste, lowers maintenance costs, promotes the sustainable development of the sponge park, and helps urban ecological construction. Description of the Drawings

[0054] Figure 1 It is a schematic flow diagram of a multi-objective collaborative optimization design method for a sponge park based on LID technology of the present invention. Detailed Embodiments

[0055] In order to make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0056] In order to better understand the above technical solutions, the above technical solutions will be described in detail below in combination with the drawings of the specification and the detailed embodiments.

[0057] As Figure 1 shown, the embodiment of the present invention provides a multi-objective collaborative optimization design method for a sponge park based on LID technology, including the following steps:

[0058] S1. Data collection and processing: Obtain the geospatial data, hydrometeorological data, ecological environment data, and socioeconomic data of the target area, and construct a parametric database of LID facilities;

[0059] S2. Model construction: Establish a multi-objective collaborative optimization model including hydrological benefits, ecological benefits, economic benefits, and social benefits;

[0060] S3. Algorithm solution: Use a hybrid intelligent optimization algorithm to solve the multi-objective collaborative optimization model and generate a Pareto optimal solution set;

[0061] S4. Solution verification: Screen the comprehensive optimal solution through coupled numerical simulation and multi-dimensional evaluation system;

[0062] S5. Dynamic optimization: Deploy an Internet of Things monitoring network and establish a digital twin system to achieve adaptive dynamic optimization of the design solution.

[0063] In this embodiment, the role of step S1 data collection and processing is to obtain multi-dimensional data required for constructing a parametric database of LID facilities, and process the data to provide an accurate and comprehensive data basis for subsequent model construction and optimization design; the specific steps are as follows:

[0064] Step S1-1: Data collection:

[0065] Geospatial data collection: Use technologies such as high-resolution remote sensing images and unmanned aerial vehicle mapping to obtain a digital elevation model (DEM) with a resolution ≤ 0.5 m in the target area to accurately present the regional topography and geomorphology; draw a spatial distribution map of soil permeability coefficient with a sampling interval ≤ 10 m through a combination of soil sampling and laboratory analysis; collect historical rainfall data from local meteorological departments to construct a rainfall intensity matrix of historical rainstorm events and record the rainfall intensity changes at different durations and different return periods;

[0066] Hydrometeorological data collection: Deploy meteorological stations and hydrological monitoring equipment in and around the target area to collect meteorological data such as rainfall, evaporation, temperature, wind speed, and wind direction in real time; monitor hydrological data such as river water level, flow rate, and flow velocity to obtain regional hydrological cycle characteristic information;

[0067] Ecological environment data collection: Conduct vegetation surveys, count different plant species and their distribution ranges, and calculate the vegetation coverage rate; Through biodiversity surveys, record the species, quantities, and habitats of animals and plants, and evaluate the biodiversity index; Combine carbon flux monitoring equipment and related models to estimate the regional carbon sequestration capacity; At the same time, collect data such as soil types, vegetation types, and ecologically sensitive areas to clarify the current status of the regional ecological environment;

[0068] Socio-economic data collection: Collect data such as the regional population quantity, distribution, economic development level, land use type, building density, etc., to understand the needs and impacts of regional socio-economic activities on the construction of sponge parks, and provide a basis for subsequent economic and social benefit evaluations;

[0069] Step S1-2: Data preprocessing:

[0070] Perform preprocessing operations on various collected data, such as format conversion, missing value filling, and outlier removal; For geospatial data, unify the coordinate system and data format to ensure the spatial consistency of the data; For hydrometeorological data, use methods such as time series interpolation to process missing values; For socio-economic data, ensure the accuracy and integrity of the data through data verification and cross-checking;

[0071] Step S1-3: Construct a parametric database for LID facilities:

[0072] Classify and integrate the preprocessed data according to the LID facility types to establish a parametric database; For permeable pavements, in addition to recording their initial porosity Determine the dynamic correction parameters (value range 0.15 - 0.25) according to factors such as their material properties and service life, for calculating the effective porosity under different operation days The calculation formula is (where, is the effective porosity of the permeable pavement when the facility operates days, is the initial porosity of the permeable pavement, is the operation days of the facility, is the dynamic correction coefficient of the porosity of the permeable pavement changing with time, reflecting the degree of porosity decrease caused by factors such as use and wear); For other LID facilities, such as rain gardens and ecological wetlands, collect their structural dimensions, infiltration performance, water storage capacity and other parameters, and establish corresponding parametric dynamic correction models to reflect the performance changes of the facilities under different operation conditions, and realize the dynamic update and maintenance of the database.

[0073] In this embodiment, the function of step S2 model construction is to establish a multi-objective collaborative optimization model covering hydrological benefits, ecological benefits, economic benefits and social benefits, providing a quantitative analysis framework for the comprehensive optimization design of sponge parks. The specific steps are as follows:

[0074] Step S2-1: Determine the multi-objective system:

[0075] Hydrological benefit objective: Take the runoff reduction rate rainwater retention time peak flow ratio as the core indicators. The runoff reduction rate reflects the ability of the sponge park to reduce rainfall runoff; the rainwater retention time reflects the retention duration of rainwater in the park, delaying the external discharge of rainwater; the peak flow ratio represents the ratio of the current peak flow to the initial peak flow, used to measure the reduction effect of sponge facilities on flood peaks;

[0076] Ecological benefit objective: Select the vegetation coverage rate biodiversity index carbon sequestration capacity as the key indicators. The vegetation coverage rate measures the degree of plant coverage in the park, affecting the stability of the ecosystem; the biodiversity index evaluates the richness of biological species and the state of ecological balance in the park; the carbon sequestration capacity reflects the ability of the park to absorb carbon dioxide, reflecting its role in climate regulation;

[0077] Economic benefit objective: Comprehensively consider factors such as construction cost, maintenance cost, and operation income, establish an economic indicator system, and evaluate the economic feasibility of the construction and operation of sponge parks;

[0078] Social benefit objective: Cover aspects such as public satisfaction, leisure and entertainment functions, and urban image improvement. Determine relevant evaluation indicators through methods such as questionnaires and social research to measure the positive impact of sponge parks on social development;

[0079] Step S2-2: Quantify each objective function:

[0080] Quantification of hydrological benefit objective: Construct a hydrological benefit objective function

[0081] (where is the quantification value of the hydrological benefit objective, used to comprehensively evaluate the hydrological benefits of the sponge park; , , are the weight coefficients corresponding to the runoff reduction rate, rainwater retention time, and peak flow ratio respectively, and satisfy , reflecting the relative importance of each indicator in the hydrological benefit evaluation; is the runoff reduction rate, calculated by the difference between the actual runoff and the runoff without sponge facilities; is the rainwater retention time, determined by the time interval from when rainwater enters the park to when it drains out; is the current peak flow rate, is the initial peak flow rate, both obtained through hydrological monitoring or simulation calculations);

[0082] Quantification of ecological benefit objectives: Establish an ecological benefit objective function (wherein, is the quantification value of ecological benefit objectives, used to comprehensively evaluate the ecological benefits of a sponge park; , , are the weight coefficients corresponding to the vegetation coverage rate, biodiversity index, and carbon sequestration capacity respectively, and satisfy , reflecting the weights of each index in the ecological benefit assessment; is the vegetation coverage rate, calculated by the ratio of the vegetation area to the total area of the park; is the biodiversity index, calculated using methods such as the Shannon - Wiener index; is the carbon sequestration capacity, estimated through carbon flux monitoring data and related models);

[0083] Step S2 - 3: Construct a multi - objective collaborative optimization model:

[0084] Integrate the above - quantified hydrological benefit, ecological benefit, economic benefit, and social benefit objective functions to construct a multi - objective collaborative optimization model; clarify the mutual relationships and constraint conditions among the objective functions in the model to ensure that the model can comprehensively and accurately reflect the requirements of multi - objective collaborative optimization in sponge park design; at the same time, consider the limiting factors in actual projects, such as land resource constraints, construction cost budgets, technical feasibility, etc., and use them as the constraint conditions of the model to limit the optimization solution, so that the solution output by the model has practical engineering application value.

[0085] In this embodiment, the role of the algorithm solution in Step S3 is to use a hybrid intelligent optimization algorithm to solve the multi - objective collaborative optimization model constructed in Step S2, obtain the Pareto optimal solution set that meets the multi - objective requirements, and provide diverse choices for subsequent scheme decision - making; the specific steps are as follows:

[0086] Step S3 - 1: Initialize algorithm parameters:

[0087] Parameter settings for the improved NSGA - II algorithm: Set the maximum number of iterations , used to control the termination condition of the algorithm's operation; initialize the range of crossover probability, based on the crossover probability dynamic adjustment formula (wherein, is the crossover probability, used to control the probability of individual crossover operations; is the current iteration number, reflecting the running process of the algorithm; is the maximum iteration number. As the iteration number increases, the crossover probability is dynamically changing between to balance the global search and local search capabilities), and determine the initial value of the crossover probability; set the elitist retention ratio to , which is used to select the top individuals with the entropy of crowding degree among the non-dominated solutions in each generation (the calculation formula of the entropy of crowding degree is , where is the entropy of crowding degree of the -th individual, measuring the distribution density of the individual in the solution space; is the number of objective functions; is the value of the -th individual on the -th objective function), ensuring that excellent individuals enter the next generation;

[0088] Parameter settings for enhancing the particle swarm optimization algorithm: Determine the minimum inertia weight , the maximum inertia weight and the decay constant (the velocity update strategy formula is

[0089] , where is the velocity of the -th particle on the -th dimension at the -th moment; is used to control the globality and locality of the particle search; is the current iteration number, which also indirectly represents the "moment" of particle state update; , are learning factors, usually taking values around 2, guiding the particle to move towards the individual optimal and global optimal positions; , are random numbers within the range; is the individual optimal position of the -th particle on the -th dimension; is the position of the -th particle on the -th dimension at the -th moment; is the value of the -th global optimal position selected from the elite solution set on the -th dimension), and at the same time set the learning factors , , as well as the boundary ranges of the particle position and velocity;

[0090] Step S3-2: Generate the initial population:

[0091] According to the solution space range of the problem, a certain number of individuals are randomly generated to form an initial population; for the multi-objective collaborative optimization problem of sponge parks, each individual represents a sponge park design scheme, including parameter information such as the type, layout, and scale of LID facilities, ensuring that the initial population has a certain distribution within the solution space and providing a diversity basis for subsequent searches;

[0092] Step S3-3: Iterative solution of the hybrid intelligent optimization algorithm:

[0093] Algorithm collaborative operation: The improved NSGA-II algorithm and the enhanced particle swarm optimization algorithm are run simultaneously. In each iteration, the two algorithms operate on the current population respectively; the improved NSGA-II algorithm performs global search and non-dominated solution screening through genetic operations such as selection, crossover, and mutation, combined with dynamic crossover probability adjustment and elitist retention strategy; the enhanced particle swarm optimization algorithm performs local optimization according to the velocity update strategy, adjusts the particle search range using the non-linear inertia term, and combines individual best and global best information;

[0094] Application of the multi-objective collaborative mechanism: In the process of algorithm solution, a multi-objective collaborative mechanism is adopted; through the dynamic weight allocation function (where, is the dynamic weight of the -th iteration and the -th objective; is the adjustment parameter, and its value range is 0.8 - 1.2, which is used to adjust the sensitivity of weight change; is the -th iteration and the -th objective's normalized fitness, reflecting the quality of this objective in the current iteration) to dynamically adjust the weights of each objective, so that the algorithm focuses on different objectives at different iteration stages; represents the normalized fitness value of the -th iteration and the -th optimization objective; The adaptive penalty function method is used to handle the constraint conditions, and the penalty factor calculation formula is (where, is the penalty factor at the -th iteration; is the initial penalty factor, and its value range is ; is the current iteration number; is the maximum iteration number. As the iteration progresses, the penalty factor gradually increases, enhancing the penalty for individuals violating the constraint conditions), ensuring that the solution result meets the actual engineering constraints;

[0095] Algorithm mode switching: Introduce a multi-modal optimization strategy switching mechanism, according to the population diversity index , Convergence speed , Pareto front change rate and other indicators to automatically switch the algorithm mode; if the population diversity index , enable the global search mode (particle swarm weight ), expand the search range; if the convergence speed , enable the local enhancement mode (mutation probability ), strengthen the local fine search; if the Pareto front change rate , trigger chaotic perturbation (where is the new position generated after chaotic perturbation; is the original position; is the chaotic parameter, and the value range is ), to avoid the algorithm falling into local optimum;

[0096] Step S3-4: Generate the Pareto optimal solution set:

[0097] After multiple rounds of iteration, when the termination conditions are met (such as reaching the maximum number of iterations, the Pareto front converges stably, etc.), the algorithm stops running; the set of non-dominated solutions obtained at this time is the Pareto optimal solution set, and no other solution in the multi-objective space can simultaneously outperform each solution in the set. These solutions cover a variety of sponge park design schemes under different objective emphases, providing rich choices for subsequent scheme verification and decision-making.

[0098] In this embodiment, the role of step S4 scheme verification is to screen out the comprehensive optimal scheme from the Pareto optimal solution set through the coupling of numerical simulation and multi-dimensional evaluation system, ensuring that the selected scheme has good performance and feasibility in practical applications; the specific steps are as follows:

[0099] Step S4-1: Numerical simulation preparation:

[0100] Data input: Collect and organize the data related to the sponge park design scheme, including geospatial data (such as terrain, land use type, etc.), hydro-meteorological data (such as rainfall, evaporation, etc.), and LID facility parameters (such as permeability coefficient, water storage capacity, etc.), and input these data into the numerical simulation model;

[0101] Model setup: Use the coupling technology of the SWMM (Storm Water Management Model) model and GIS (Geographic Information System) spatial analysis for numerical simulation; in the SWMM model, set parameters such as sub-catchments, pipe network systems, and LID facilities according to the actual situation to determine the simulation time step and simulation period; at the same time, use the GIS spatial analysis function to process and analyze geographical spatial data to provide accurate spatial information for the SWMM model;

[0102] Step S4-2: Runoff numerical simulation:

[0103] Simulation calculation: According to the settings in Step S4-1, run the SWMM model for runoff simulation calculation; the runoff calculation satisfies the formula

[0104] (where, is the total runoff, which is the target value to be simulated and calculated; is the number of sub-catchments, reflecting the division of the study area; is the th sub-catchment area, which can be obtained through GIS spatial analysis; is the calculation period, determined according to actual needs; is the th sub-catchment's rainfall intensity at time , which can be obtained from hydrometeorological data; is the th sub-catchment's dynamic infiltration rate function at time , which is related to factors such as soil type and antecedent soil moisture; is the th sub-catchment's attenuation coefficient, which affects the change of infiltration rate over time; represents the differential with respect to time and is used in the integral operation to infinitely subdivide the time period [0, T] for accurate calculation of runoff);

[0105] Result output: After the simulation ends, output the runoff of each sub-catchment at different times, as well as the change curve of the total runoff of the entire study area over time; through the analysis of the simulation results, the effects of different design schemes on runoff control can be evaluated;

[0106] Step S4-3: Construction of a multi-dimensional evaluation system:

[0107] Index selection: Evaluation indicators are selected from four aspects: hydrological benefits, ecological benefits, economic benefits, and social benefits; hydrological benefit indicators include runoff reduction rate, peak flow reduction rate, etc.; ecological benefit indicators include vegetation coverage rate, biodiversity index, etc.; economic benefit indicators include construction cost, maintenance cost, etc.; social benefit indicators include public satisfaction, recreational function, etc.

[0108] Index quantification: Each evaluation indicator is quantified, and different types of indicators are converted into comparable values; for example, the runoff reduction rate can be calculated by the runoff volume obtained from simulation and the runoff volume without LID facilities; public satisfaction can be quantified through questionnaire surveys.

[0109] Weight determination: Methods such as the Analytic Hierarchy Process (AHP) and entropy weight method are used to determine the weights of each evaluation indicator, reflecting the relative importance of each indicator in the comprehensive evaluation.

[0110] Step S4-4: Scheme evaluation and screening:

[0111] Calculation of closeness degree: The improved

[0112] TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) method is used to evaluate each design scheme; first, a decision matrix is constructed based on the quantified values and weights of each evaluation indicator; then, the Mahalanobis distances between each scheme and the positive ideal solution and the negative ideal solution are calculated and ; finally, according to the closeness degree calculation formula ( ) ( is the closeness degree of the th scheme, used to measure the closeness of the scheme to the ideal solution; is the Mahalanobis distance between the th scheme and the positive ideal solution, reflecting the gap between the scheme and the optimal scheme; is the Mahalanobis distance between the th scheme and the negative ideal solution, reflecting the gap between the scheme and the worst scheme; is the adjustment coefficient, and its value range is , used to balance and in the calculation of the closeness degree) to calculate the closeness degree of each scheme;

[0113] Scheme screening: All design schemes are sorted according to the size of the closeness degree, and the larger the closeness degree, the better the scheme; the scheme with the largest closeness degree is selected as the comprehensive optimal scheme, and this scheme has a better comprehensive performance in terms of hydrological benefits, ecological benefits, economic benefits, and social benefits.

[0114] Step S4-5: Result verification and feedback:

[0115] Result verification: Further verify the comprehensively optimal solution selected. This can be done through on-site research, comparison with historical data, etc., to check the degree of conformity between the simulation results and the actual situation. If a large deviation is found between the simulation results and the actual situation, it is necessary to re-check links such as data input and model settings and correct the simulation process.

[0116] Feedback optimization: Feed back the verification results to the model construction in step S2 and the algorithm solving process in step S3, and optimize and adjust the model parameters and algorithms to improve the accuracy of subsequent solution design and evaluation.

[0117] In this embodiment, the role of the dynamic optimization in step S5 is to achieve the adaptive update of the sponge park design solution by deploying an Internet of Things monitoring network and establishing a digital twin system, ensuring that the park can maintain good performance and benefits under different environmental conditions and operation stages. The specific steps are as follows:

[0118] Step S5-1: Deployment of the Internet of Things monitoring network:

[0119] Deployment of distributed hydrological sensing units:

[0120] Laser disdrometer: Reasonably arrange laser disdrometers in different areas of the sponge park, and its accuracy needs to reach ±5%. These instruments can monitor parameters such as the size, speed, and quantity of raindrops in real time, so as to accurately obtain the microscopic characteristic information of rainfall and provide basic data for subsequent runoff analysis and prediction.

[0121] Capacitive soil moisture sensor: Install capacitive soil moisture sensors at an appropriate density, and its measurement range is 0-100%. The sensors need to be evenly distributed in different soil depths and different vegetation coverage areas of the park to comprehensively and accurately monitor the changes in soil moisture and reflect the water storage capacity and water dynamics of the soil.

[0122] Ultrasonic water level gauge: Install ultrasonic water level gauges in the water bodies (such as ponds, ditches, etc.) in the park and at key drainage nodes, and the resolution should reach 1mm. It is used to monitor the changes in water level in real time and promptly grasp the water accumulation situation and water flow state in the park.

[0123] Setting of edge computing nodes:

[0124] Built-in LSTM prediction model: Build a long short-term memory (LSTM) prediction model in the edge computing node. This model is trained using historical monitoring data and can calculate the runoff coefficient deviation value in real time based on the current monitoring information , and the calculation formula is (where is the runoff coefficient deviation value, reflecting the degree of difference between the actual runoff situation and the design expectation; is the measured runoff coefficient, calculated from the monitoring data of the distributed hydrological sensing unit; is the designed runoff coefficient, which is the target value determined in the design stage of the sponge park);

[0125] Real-time data analysis and processing: The edge computing node performs real-time analysis and processing on the data collected by the distributed hydrological sensing unit. In addition to calculating the runoff coefficient deviation value, it can also perform preliminary filtering, correction, etc. on the data to reduce the data transmission volume and error, and improve the quality and usability of the data;

[0126] Data transmission module construction:

[0127] LoRaWAN protocol support: Adopt a data transmission module that supports the LoRaWAN protocol to achieve reliable data transmission between the distributed hydrological sensing unit and the edge computing node; This protocol has the advantages of low power consumption, long distance, large capacity, etc., and can meet the requirements of large-area monitoring and long-term stable operation of the sponge park;

[0128] Packet loss rate control: By optimizing the network topology structure, adjusting transmission parameters, etc., ensure that the packet loss rate of data transmission ≤ 0.1%, and ensure that the monitoring data can be accurately and timely transmitted to the digital twin system;

[0129] Step S5-2: Digital twin system establishment:

[0130] Physical entity layer construction:

[0131] Deployment of LID facility operation status sensors: Deploy LID facility operation status sensors in the sponge park at a density of ≥ 3 per 100 m², such as monitoring the clogging situation of permeable pavement, the water level change of rain gardens, the water quality indicators of ecological wetlands, etc.; These sensors can obtain the operation status information of LID facilities in real time and provide accurate physical entity data for the digital twin system;

[0132] Virtual model layer integration:

[0133] Integration of SWMM-HEC coupling simulation engine: Integrate the SWMM-HEC coupling simulation engine into the virtual model layer, and the time step of this engine needs to be ≤ 1 minute; The SWMM model is used to simulate the hydrological and hydraulic processes of the urban rainwater system, and the HEC model can conduct a more in-depth analysis of river hydrology and water conservancy projects; Through the coupling of the two, it can more accurately simulate the water flow movement, water quality change and the performance of LID facilities in the sponge park under different rainfall conditions;

[0134] Virtual model construction and calibration: Construct a virtual model based on the actual geospatial data, LID facility parameters, and monitoring data of the sponge park; calibrate and validate the virtual model using historical monitoring data to ensure that the virtual model can accurately reflect the operating status and performance of the physical entity;

[0135] Implementation of the data interaction layer:

[0136] Adoption of the OPCUA protocol: Adopt the OPCUA protocol to achieve real-time data synchronization between the physical entity layer and the virtual model layer, ensuring that the data transmission latency ≤ 200ms; the OPCUA protocol has the advantages of strong openness, security, and interoperability, enabling seamless data exchange between different devices and systems;

[0137] Establishment of a data synchronization mechanism: Establish a sound data synchronization mechanism to ensure that the real-time monitoring data of the physical entity layer can be updated to the virtual model layer in a timely and accurate manner, and at the same time, the analysis results and decision-making instructions of the virtual model layer can also be fed back to the physical entity layer to achieve two-way interaction between the physical entity and the virtual model;

[0138] Embedding of the decision optimization layer:

[0139] Embedding of the mixed integer programming algorithm: Embed the mixed integer programming algorithm in the decision optimization layer, and combine the real-time data obtained from the IoT monitoring network and the simulation analysis results of the virtual model layer to optimize the operation and maintenance costs of the sponge park; for example, reasonably arrange the irrigation plan according to soil moisture and rainfall prediction information; formulate the best maintenance and update strategies according to the operating status and performance indicators of LID facilities;

[0140] Step S5-3: Dynamic optimization trigger and response:

[0141] Monitoring of trigger conditions:

[0142] Real-time monitoring and judgment: Use the IoT monitoring network to real-time monitor key indicators such as the runoff coefficient deviation value , vegetation coverage rate, and rainstorm recurrence period; when the following dynamic optimization trigger conditions are met, the system automatically starts the optimization program:

[0143] When For 3 consecutive sampling periods, it indicates that the actual runoff situation deviates significantly from the design expectation, which may be due to the decline in the performance of LID facilities or abnormal rainfall conditions;

[0144] The vegetation coverage rate drops by more than 20% of the design value, indicating that the ecological environment of the park may be damaged, affecting the ecological benefits;

[0145] The rainstorm recurrence period exceeds 1.3 times the design standard, meaning that the current rainfall intensity exceeds the design tolerance of the park, and it is necessary to adjust the response strategy in a timely manner;

[0146] Optimization Response Implementation:

[0147] Regular Optimization and Emergency Optimization: Different optimization response measures are taken according to the severity and urgency of the triggering conditions;

[0148] Regular Optimization: When the triggering condition is relatively mild, a regular optimization cycle is started, and the optimization cycle ≤ 48 hours; at this stage, the decision-making optimization layer uses a mixed-integer programming algorithm, combines real-time data and virtual model analysis, and formulates optimization plans, such as adjusting the operating parameters of LID facilities, arranging appropriate maintenance work, etc.;

[0149] Emergency Optimization: When the triggering condition is relatively severe, an emergency optimization cycle is started, and the emergency optimization cycle ≤ 6 hours; at this time, the system quickly takes emergency measures, such as increasing the operating power of drainage equipment, temporarily adjusting the use function of the park, etc., to ensure the safety and normal operation of the park;

[0150] Optimization Effect Evaluation and Feedback:

[0151] Effect Evaluation: After the optimization measures are implemented for a period of time, the optimization effect is evaluated; by comparing various performance indicators before and after optimization (such as runoff control effect, ecological benefit, economic benefit, etc.), it is judged whether the optimization measures are effective;

[0152] Feedback Adjustment: The optimization effect evaluation results are fed back into the digital twin system to update and adjust the virtual model. At the same time, summarize experience and lessons, provide reference for subsequent dynamic optimization, and continuously improve the adaptive ability and operation efficiency of the sponge park.

[0153] Furthermore, the Internet of Things monitoring network is the key infrastructure for realizing the dynamic optimization of the sponge park. Through the collaborative work of multi-type sensing devices, intelligent computing nodes and efficient data transmission modules, it provides real-time and accurate data support for the full-life cycle digital twin system, including the following components and functions:

[0154] I. Distributed Hydrological Sensor Array:

[0155] Laser Disdrometer: Deployed in different functional areas of the sponge park, with high-precision monitoring ability of ±5%; its core function is to capture in real time the microscopic rainfall characteristic data such as the particle size distribution, falling speed and rainfall per unit time of raindrops; through the analysis of these data, parameters such as rainfall intensity and rainfall kinetic energy can be accurately calculated, providing basic data support for predicting runoff generation and evaluating the rainwater interception effect of LID facilities, and then optimizing the rainwater management strategy of the park;

[0156] Capacitive soil moisture sensors: Installed according to different vegetation types and soil textures in different zones, with a measurement range covering 0 - 100%; These sensors can dynamically monitor changes in soil water content, timely feedback on the soil water storage status, and provide key data for the intelligent control of irrigation systems, the assessment of vegetation growth health, and the analysis of rainwater infiltration capacity; For example, when the soil moisture is close to saturation, it can warn of possible surface runoff risks and assist in adjusting the operation parameters of LID facilities;

[0157] Ultrasonic water level gauges: Installed at key water level monitoring points such as artificial lakes, rainwater storage ponds, and drainage ditches in the park, with a resolution of 1 mm; By continuously monitoring water level changes, it can grasp the water storage and drainage conditions of the water bodies inside the park in real time. Combining rainfall data and the operation status of LID facilities, it can predict the risk of waterlogging and provide real-time data basis for drainage system scheduling and emergency response decision-making;

[0158] II. Edge computing nodes:

[0159] Built-in LSTM prediction model: Built-in a long short-term memory (LSTM) prediction model, which is trained based on historical hydrometeorological data and LID facility operation data; Its core task is to process the data collected by distributed sensing units in real time, and through complex time series data analysis, calculate the runoff coefficient deviation value (where, is the runoff coefficient deviation value, reflecting the degree of difference between the actual runoff situation and the design expectation; is the measured runoff coefficient, calculated from the monitoring data of the distributed hydrological sensing unit; is the designed runoff coefficient, which is the target value determined during the design stage of the sponge park); This deviation value can intuitively reflect the difference between the actual hydrological conditions of the park and the design expectation, providing a quantitative basis for dynamic optimization;

[0160] Real-time data preprocessing: Perform real-time preprocessing operations on sensing data such as filtering, denoising, outlier detection, and correction; Through algorithms, remove interference signals during data collection, ensure the authenticity and reliability of the data transmitted to the digital twin system, reduce the data transmission volume, improve data processing efficiency, and reduce the network load pressure;

[0161] III. Data transmission module:

[0162] Support for LoRaWAN protocol: Adopt the Low Power Wide Area Network (LoRaWAN) protocol as the data transmission standard, with the characteristics of long-distance transmission (up to several kilometers), low power consumption (the device battery life can be up to several years), and large-capacity access (a single gateway can support thousands of nodes); This protocol is especially suitable for the data transmission requirements of large-area and decentralized sensing devices in sponge parks and can achieve stable data communication in complex environments;

[0163] High-reliability transmission guarantee: By optimizing technical means such as network topology structures (such as a hybrid network of star and Mesh types), dynamically adjusting transmission power and channels, the data packet loss rate is strictly controlled at ≤0.1%; at the same time, data encryption transmission technology is adopted to ensure the security and integrity of monitoring data during the transmission process, prevent data leakage or tampering, and ensure that the information obtained by the digital twin system is accurate and error-free;

[0164] Through the close cooperation of each module, the IoT monitoring network constructs a complete chain from data collection, processing to transmission, providing a solid technical foundation for the real-time monitoring, intelligent analysis and dynamic optimization of the sponge park, and realizing the efficient data interaction between the physical entity and the digital virtual space.

[0165] Furthermore, relying on the IoT monitoring network and the full-life-cycle digital twin system, the adaptive dynamic optimization realizes the dynamic adjustment of the sponge park design scheme in a way driven by real-time data and decision-making by intelligent algorithms, ensuring that it continuously maintains the multi-objective collaborative optimal state under complex environmental changes. The specific operation mechanism is as follows:

[0166] I. Dynamic optimization trigger mechanism:

[0167] Runoff coefficient deviation trigger: Rainfall and runoff data are monitored in real time through devices such as laser disdrometers and ultrasonic water level gauges in the IoT monitoring network, and the runoff coefficient deviation value is calculated by the LSTM prediction model of the edge computing node ; when and it lasts for 3 sampling periods, it indicates that the actual runoff situation significantly deviates from the design expectation, and there may be problems such as blockage of LID facilities and decline in soil infiltration capacity, triggering the dynamic optimization program;

[0168] Ecological index anomaly trigger: The change of vegetation coverage in the park is monitored in real time by using a vegetation coverage sensor. When the vegetation coverage decreases by more than 20% of the design value, it means that the park's ecological system is threatened, which may affect ecological benefits such as biodiversity and carbon sequestration capacity, and then trigger the optimization process to take remedial measures such as vegetation replanting and irrigation adjustment;

[0169] Extreme rainfall event trigger: According to meteorological monitoring data and historical rainfall models, when the rainstorm recurrence period exceeds 1.3 times the design standard, it indicates that the current rainfall intensity exceeds the original design coping capacity of the park. To avoid the risks of waterlogging and facility damage, the system automatically starts dynamic optimization and quickly adjusts the drainage strategy and activates emergency storage facilities;

[0170] II. Optimization response grading mechanism:

[0171] Regular optimization response: For non-urgent minor deviations or slow changes (such as the gradual decline in the performance of some LID facilities, seasonal vegetation growth changes, etc.), initiate the regular optimization process with a response cycle ≤ 48 hours; during this period, the decision-making optimization layer of the digital twin system uses the mixed-integer programming algorithm to comprehensively consider multiple objectives such as hydrology, ecology, and economy, and formulates optimization plans, such as adjusting the operating parameters of LID facilities (such as porosity correction of permeable pavements, water level control of rain gardens), arranging planned maintenance work (such as cleaning blocked stormwater pipes, pruning vegetation);

[0172] Emergency optimization response: When encountering emergencies (such as the risk of waterlogging caused by sudden heavy rainfall, sudden equipment failures, etc.), trigger the emergency optimization process with a response cycle ≤ 6 hours; the system gives priority to ensuring the safe operation of the park and quickly takes emergency measures, such as starting backup drainage pumps, temporarily closing some areas, remotely controlling the emergency drainage of LID facilities, etc.; at the same time, quickly simulate the effects of different emergency plans through the digital twin system to assist in making the optimal disposal strategy;

[0173] III. Optimization strategy generation and implementation:

[0174] Data-driven intelligent decision-making: The virtual model layer of the digital twin system is based on the real-time data of the Internet of Things monitoring network, combined with the SWMM-HEC coupling simulation engine, to dynamically simulate the operation status of the sponge park; by comparing the multi-objective benefits (hydrological benefits, ecological benefits, etc.) of different design plans in the current environment, use the hybrid intelligent optimization algorithm (such as the improved NSGA-II algorithm, enhanced particle swarm optimization algorithm) to generate the Pareto optimal solution set, and then combine the dynamic weight allocation function , and screen out the most suitable optimization plan at present;

[0175] Precision measure implementation: According to the generated optimization plan, transmit control instructions to relevant devices in the physical entity layer through the data interaction layer (OPCUA protocol); for example, adjust intelligent valves to control the flow direction of rainwater, start the automatic irrigation system to improve the vegetation growth conditions, dispatch maintenance personnel to repair faulty facilities, etc., to achieve the precise implementation of optimization strategies;

[0176] IV. Optimization effect evaluation and continuous improvement:

[0177] Effect evaluation: After the implementation of the optimization measures, the Internet of Things monitoring network continuously collects relevant data to quantitatively evaluate the optimization effect; compare the changes in hydrological, ecological, economic and other indicators before and after optimization, such as runoff reduction rate, vegetation coverage recovery, maintenance cost changes, etc., to judge whether the optimization measures have achieved the expected goals;

[0178] Feedback and Iteration: Feed the evaluation results back to the digital twin system to calibrate the virtual model and update parameters. At the same time, summarize the optimization experience to provide reference for subsequent similar situations. If the optimization effect fails to meet the expectations, restart the optimization process, adjust the optimization strategy, and form a closed-loop adaptive dynamic optimization system of "monitoring - analysis - decision - execution - evaluation - improvement" to ensure the efficient and sustainable operation of the sponge park at all times.

[0179] In this embodiment, the digital twin system serves as the core hub for the dynamic optimization of the sponge park. By constructing a two-way mapping relationship between the physical entity and the virtual model, and combining real-time data-driven and intelligent algorithm decision-making, it realizes the digital management and optimization of all elements and processes in the park. The system consists of a physical entity layer, a virtual model layer, a data interaction layer, and a decision optimization layer. Each layer operates in coordination as follows:

[0180] I. Physical Entity Layer:

[0181] Monitoring the Operating Status of LID Facilities: Deploy multiple types of sensors on LID facilities such as permeable pavements, rain gardens, and ecological wetlands in the sponge park at a density of ≥3 sensors per 100 m². For example, embed porosity sensors and pressure sensors in the permeable pavement to monitor its clogging situation and infiltration performance in real time; install water level sensors and water quality sensors around the rain garden to obtain data on water level changes and pollutant purification effects; arrange biodiversity monitoring equipment and soil moisture sensors in the ecological wetland to grasp the health status of the ecosystem. These sensors continuously collect the operating parameters of the facilities to provide basic data of the real world for the virtual model.

[0182] Real-time Sensing of Environmental Elements: Combine devices such as laser disdrometers, capacitive soil moisture sensors, and ultrasonic water level gauges in the Internet of Things monitoring network to comprehensively sense environmental elements such as rainfall, soil, and water bodies in the park. For example, the laser disdrometer captures rainfall characteristics in real time to provide input for the virtual model to simulate the runoff process; the soil moisture sensor dynamically feedbacks the soil water content to assist in evaluating rainwater infiltration and vegetation water requirements, and constructs a real-time digital mirror of the physical environment.

[0183] II. Virtual Model Layer:

[0184] SWMM-HEC Coupled Simulation Engine Integration: Integrate SWMM (Storm Water Management Model) and HEC (Hydrologic Engineering Center Model) to perform high-precision simulations with a time step of ≤1 minute; the SWMM model is used to finely simulate the hydrological and hydraulic processes of the drainage systems such as stormwater pipelines and LID facilities in the sponge park, including runoff generation, collection, transmission, and regulation; the HEC model focuses on simulating the hydrodynamic and water quality changes of complex water bodies (such as lakes and rivers); after the two are coupled, it can completely reproduce the water flow movement, pollutant migration, and ecological response processes in the park under different rainfall conditions and facility operation states, providing a dynamic simulation environment for multi-objective optimization;

[0185] Virtual Model Construction and Calibration: Based on the geospatial data of the target area (such as high-precision DEM models and soil permeability coefficient distribution maps), the LID facility parameterized database, and historical monitoring data, construct a three-dimensional virtual model; by comparing the model simulation results with the actual monitoring data (such as measured water levels and runoff), use parameter calibration algorithms (such as genetic algorithms and particle swarm algorithms) to adjust the model parameters to ensure that the virtual model can accurately map the operating laws of physical entities, with the error controlled within the engineering acceptable range;

[0186] III. Data Interaction Layer:

[0187] Real-time Synchronization via OPCUA Protocol: Adopt the OPCUA (Open Platform Communications Unified Architecture) protocol as the data interaction standard to ensure low-latency (≤200ms) and highly reliable data transmission between the physical entity layer and the virtual model layer; this protocol supports seamless data interaction between heterogeneous systems, can quickly transmit the real-time data collected by sensors to the virtual model layer for update, and at the same time accurately issue the control instructions of the virtual model to the physical entity devices to achieve two-way data flow;

[0188] Data Cleaning and Standardization Processing: During the data transmission process, perform preprocessing operations such as cleaning, noise reduction, and format conversion on the original monitoring data; by establishing a unified data standard, convert different types and sources of data (such as sensor data and model output data) into a format that can be recognized and processed by each layer of the system to ensure data consistency and availability, and avoid model errors or decision-making mistakes caused by data differences;

[0189] IV. Decision Optimization Layer:

[0190] Embedding of Mixed Integer Programming Algorithm: For the optimization problem of the construction, operation and maintenance costs of sponge parks, a mixed integer programming algorithm is embedded; based on a multi-objective collaborative optimization model (hydrological benefits, ecological benefits, economic benefits and social benefits), combined with real-time monitoring data and virtual model prediction results, the best resource allocation plan is solved under the satisfaction of engineering constraints (such as land resource limitations, budget caps); for example, optimizing the construction scale and layout of LID facilities, formulating equipment maintenance plans, adjusting operation strategies, to achieve cost minimization and benefit maximization;

[0191] Dynamic Decision Support: Based on the simulation analysis of the virtual model and the calculation results of the optimization algorithm, provide visual decision support for managers; by generating multi-scheme comparison reports, benefit evaluation charts, risk warning information, etc., assist decision-makers to quickly understand the potential impacts of different strategies and select the optimal management plan; at the same time, the system supports the dynamic simulation verification of decision-making plans, anticipates the implementation effects in advance, reduces decision-making risks, and promotes the intelligent and scientific management of sponge parks.

[0192] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-objective collaborative optimization design method for sponge parks based on LID technology, characterized in that: It includes the following steps: S1. Data collection and processing: Obtain the geospatial data, hydrometeorological data, ecological environment data, and socioeconomic data of the target area, and construct a parametric database of LID facilities; S2. Model construction: Establish a multi-objective collaborative optimization model including hydrological benefits, ecological benefits, economic benefits, and social benefits; S3. Algorithm solution: Use a hybrid intelligent optimization algorithm to solve the multi-objective collaborative optimization model and generate a Pareto optimal solution set; The hybrid intelligent optimization algorithm includes: The improved NSGA-II algorithm, and its improvements include: Dynamically adjust the crossover probability, and the crossover probability value changes periodically with the number of iterations according to the cosine function, and the change range is from 0.5 to 0.9; An elite retention strategy based on crowding entropy, and retain the non-dominated solutions with the top 15% of the crowding entropy values in each generation; The enhanced particle swarm optimization algorithm, and its velocity update includes a non-linear decreasing inertia weight term, and the inertia weight decays from the initial value of 0.9 to 0.4 according to the exponential function; The multi-objective collaborative mechanism of the hybrid intelligent optimization algorithm includes: A dynamic weight allocation function, and the weights of each objective are dynamically adjusted according to the exponential function of the normalized fitness value, and the adjustment coefficient η is from 0.8 to 1.2; Adaptive penalty function method, the penalty factor increases exponentially with the number of iterations, and the initial penalty factor ranges from 10³ to 10 4 ; The hybrid intelligent optimization algorithm introduces a multi-modal strategy switching mechanism: When the population diversity index H > 0.7, enable the global search mode and set the particle swarm inertia weight > 0.8; When the convergence speed <0.05, enable the local enhancement mode and set the mutation probability > 0.15; When the change rate δ of the Pareto front is less than 0.01, chaotic perturbation is triggered, and the perturbation formula is , where is the new position generated after chaotic perturbation, is the original position, is the chaotic parameter, ranges from 3.8 to 4.0; S4. Scheme verification: Screen the comprehensive optimal scheme through the coupling of numerical simulation and multi-dimensional evaluation system. The coupling numerical simulation uses the SWMM model and GIS spatial analysis technology to couple and calculate the runoff, and the calculation process includes the integral operation of the dynamic infiltration rate function and the roughness coefficient correction coefficient; The multi-dimensional evaluation system uses the improved TOPSIS method. The discrimination coefficient ρ is introduced in the calculation of the closeness, and the value range is from 0.4 to 0.6, and the proximity of the scheme to the ideal solution is measured based on the Mahalanobis distance; S5. Dynamic optimization: Deploy an Internet of Things monitoring network and establish a digital twin system to realize the adaptive dynamic optimization of the design scheme.

2. A multi-objective collaborative optimization design method for a sponge park based on LID technology according to claim 1, characterized in that: In the step S1: The geospatial data includes a digital elevation model with a resolution ≤ 0.5 m, a spatial distribution map of soil permeability coefficient with a sampling interval ≤ 10 m, and a rainfall intensity matrix of historical rainstorm events; The parametric database of LID facilities includes a dynamic correction coefficient of the effective porosity of the permeable pavement, and the dynamic correction coefficient is adjusted annually according to the β coefficient according to the number of operating days. Among them, the value range of β is from 0.15 to 0.

25.

3. A multi-objective collaborative optimization design method for a sponge park based on LID technology according to claim 1, characterized in that: In the step S2, the multi-objective collaborative optimization model satisfies: The hydrological benefit target is quantified by the weighted sum of the surface runoff reduction rate, rainwater retention time, and peak flow delay coefficient, and the weight coefficients α1, α2, α3 satisfy α1 + α2 + α3 = 1; The ecological benefit target is quantified by the weighted sum of the vegetation coverage rate, biodiversity index, and carbon sequestration capacity, and the weight coefficients β1, β2, β3 satisfy β1 + β2 + β3 = 1.

4. A multi-objective collaborative optimization design method for a sponge park based on LID technology according to claim 1, characterized in that: The Internet of Things monitoring network in the step S5 includes: Distributed hydrological sensor array, including a laser raindrop spectrometer with an accuracy of ±5%, a capacitive soil moisture sensor with a range of 0-100%, and an ultrasonic water level gauge with a resolution of 1 mm; Edge computing node, built-in long short-term memory model for real-time calculation of runoff coefficient deviation value , The calculation formula is the absolute percentage deviation between the measured value and the designed value; Data transmission module, supporting the LoRaWAN protocol and with a packet loss rate ≤0.1%.

5. A multi-objective collaborative optimization design method for a sponge park based on the LID technology according to claim 4, characterized in that: The triggering conditions for the adaptive dynamic optimization are as follows: Runoff coefficient deviation value Exceeding 15% for three consecutive sampling periods; The vegetation coverage rate drops by more than 20% compared with the designed value; The rainstorm recurrence period exceeds 1.3 times the design standard; The optimization response time meets the requirements of ≤48 hours in the normal mode and ≤6 hours in the emergency mode.

6. A multi-objective collaborative optimization design method for a sponge park based on LID technology according to claim 1, characterized in that: The digital twin system includes: Physical entity layer, deploying LID facility status sensors with a density ≥3 per 100 m²; Virtual model layer, integrating a SWMM-HEC coupled simulation engine with a time step ≤1 minute; Data interaction layer, achieving real-time synchronization with a delay ≤200 ms using the OPC UA protocol; Decision optimization layer, embedding a mixed integer programming algorithm for full life cycle cost optimization.

Images