A method for collaborative restoration of a mine water ecological water supplementing wetland in a subsidence water accumulation area

Abstract

The present application belongs to the technical field of wastewater treatment, and particularly relates to a subsidence water area mine water ecological water supplement wetland collaborative repair method, comprising the following steps: step 1: dividing the coal mining subsidence water area into a plurality of subsidence water units according to the ground elevation boundary, and recording the total flow of mine water outflow; step 2: after completing the local search of all frog groups, mixing all the frogs, re-executing the Pareto sorting and grouping, entering the next round of iteration, and selecting the target configuration scheme from the Pareto front set after the iteration is terminated; step 3: constructing the artificial wetland in each subsidence water unit according to the target configuration scheme, planting plants and the water inlet position of the determined matrix type, and completing the subsidence water area mine water ecological water supplement wetland collaborative repair. The present application overcomes the prior art which ignores the hydraulic connection between units, lacks the multi-target global optimization capability, and does not consider the influence of the field environmental factors on the removal effect.

Description

Technical Field

[0001] This invention belongs to the field of wastewater treatment technology, specifically relating to a method for the coordinated restoration of mine water ecological replenishment wetlands in subsidence and waterlogged areas. Background Technology

[0002] When applying constructed wetlands to mine water ecological replenishment and restoration scenarios in coal mining subsidence and water accumulation areas, existing technologies still face the following problems.

[0003] First, existing solutions lack a systematic consideration of the spatial heterogeneity of subsidence and waterlogging areas. Coal mining subsidence areas typically consist of multiple waterlogged depressions with varying surface elevations, areas, and depths. These depressions are hydraulically connected due to elevation differences. When the elevation difference between adjacent units exceeds a certain threshold, water from the upstream unit flows into the downstream unit as runoff, forming a series of water quality transfer chains. Existing wetland design methods usually treat each waterlogged area as an independent unit, failing to establish a calculation model for water quality transfer between these series of units. This leads to an underestimation or overestimation of the influent water quality in the downstream unit, resulting in a discrepancy between the design and actual operational performance.

[0004] Second, existing solutions rely primarily on engineering experience or single-factor experimental methods to determine key decision variables such as substrate type, plant species, inlet layout, and mine water distribution ratio, lacking systematic optimization methods for multiple water quality objectives. In practical engineering, ammonia nitrogen removal and total phosphorus removal efficiency often cannot be simultaneously optimized. For example, selecting zeolite substrates with the strongest ammonia nitrogen adsorption capacity may result in unsatisfactory total phosphorus removal, while selecting biochar substrates with strong total phosphorus fixation capacity may lead to a decrease in ammonia nitrogen removal efficiency. Existing methods struggle to achieve a global balance among multiple interdependent water quality objectives and cannot simultaneously maximize the number of compliant units in each water collection area.

[0005] Third, existing technologies do not adequately consider the impact of on-site environmental factors on removal efficiency. Studies have shown that when dissolved oxygen concentration in water is below a certain threshold, nitrification by rhizosphere microorganisms is inhibited, and ammonia nitrogen removal efficiency decreases significantly. When water depth exceeds a certain limit, the root activity and photosynthetic efficiency of emergent plants decrease, and the plants' ability to absorb and convert nitrogen and phosphorus diminishes. Existing wetland design methods typically use standard removal coefficients obtained under laboratory conditions for calculation, without incorporating correction mechanisms for on-site factors such as dissolved oxygen and water depth, resulting in a significant deviation between theoretical predictions and actual effluent concentrations.

[0006] Fourth, existing solutions are mostly one-off designs, lacking a dynamic feedback and update mechanism during operation. The actual operating effect of constructed wetlands is affected by factors such as seasonal changes, fluctuations in hydrological conditions, substrate aging, and plant growth cycles. The initial design may no longer be suitable for actual operating conditions after a period of operation. Existing technologies lack a closed-loop control strategy that automatically triggers scheme adjustments based on operational monitoring data. When the actual effluent concentration of a certain unit consistently exceeds expectations, engineers need to rely on manual judgment to adjust the scheme, resulting in a delayed response and a lack of systematic approach. Summary of the Invention

[0007] Therefore, the main objective of this invention is to provide a method for the coordinated restoration of mine water ecological replenishment wetlands in subsidence and waterlogged areas. This method overcomes the shortcomings of existing technologies, such as ignoring the hydraulic connection between units, lacking multi-objective global optimization capabilities, not considering the impact of on-site environmental factors on the removal effect, and lacking a dynamic feedback mechanism during operation. It achieves efficient coordination between the resource utilization of mine water and the ecological restoration of subsidence and waterlogged areas.

[0008] The technical solution adopted in this invention is as follows:

[0009] A method for the coordinated restoration of mine water ecological replenishment wetlands in subsidence and waterlogged areas includes the following steps:

[0010] Step 1: Divide the coal mining subsidence and water accumulation area into several subsidence and water accumulation units according to the surface elevation boundary, collect the water quality and hydrological parameters of each subsidence and water accumulation unit, record the relationship between the surface elevation difference and surface runoff direction between adjacent subsidence and water accumulation units, and record the total flow rate of mine water outflow.

[0011] Step 2: For each subsidence water accumulation unit, determine the decision variables including substrate type, planted vegetation, inlet location, and mine water allocation ratio. Concatenate the decision variables of all subsidence water accumulation units into a decision vector and define it as one frog. Randomly generate several frogs to form an initial population. The sum of the mine water allocation ratios of all subsidence water accumulation units in each frog is equal to 100%. Based on the decision vector of each frog and the field information collected in Step 1, calculate the comprehensive removal coefficient for each unit and calculate multiple optimization target values ​​corresponding to each frog accordingly. Perform Pareto sorting on all frogs in the population and distribute them to several frog groups at equal intervals. Within each frog group, update the decision vector of the worst frog in the local group based on the best frog in the local group and determine whether to replace it based on the Pareto dominance relationship. After completing the local search of all frog groups, mix all frogs, re-execute Pareto sorting and grouping, and enter the next iteration. After the iteration terminates, select the target configuration scheme from the Pareto front set.

[0012] Step 3: Construct artificial wetlands in each subsidence and water accumulation unit according to the target configuration scheme, based on the determined substrate type, plant species, and inlet location. Continuously pump mine water into each subsidence and water accumulation unit according to the determined mine water distribution ratio. After purification by the artificial wetlands, the mine water is replenished into the water body of the subsidence and water accumulation unit, thus completing the ecological replenishment wetland collaborative restoration of mine water in the subsidence and water accumulation area.

[0013] Furthermore, in step 1, the water quality parameters collected for each subsidence water accumulation unit include dissolved oxygen concentration, ammonia nitrogen concentration, and total phosphorus concentration, while the hydrological parameters include water accumulation area and average water accumulation depth.

[0014] Furthermore, in step 2, the substrate type is selected from zeolite, gravel, and biochar; the planting plant is selected from reed, cattail, and calamus; the inlet location is selected from one end of the long side, the middle of the long side, and the four corners; the mine water distribution ratio is the percentage of the mine water flow allocated to the corresponding subsidence water accumulation unit to the total mine water outflow, with a step size of 5% and a range of 5% to 60%.

[0015] Furthermore, in step 2, the method for calculating the comprehensive removal coefficient unit by unit is as follows: the matrix ammonia nitrogen removal coefficient is determined by matrix type, with zeolite taking 0.60, biochar taking 0.50, and gravel taking 0.30; the matrix total phosphorus removal coefficient is determined by matrix type, with biochar taking 0.60, zeolite taking 0.40, and gravel taking 0.20; the plant ammonia nitrogen removal coefficient is determined by plant species, with reed taking 0.30, cattail taking 0.20, and calamus taking 0.15; the plant total phosphorus removal coefficient is determined by plant species, with cattail taking 0.30, reed taking 0.20, and calamus taking 0.15; the matrix ammonia nitrogen removal coefficient and the plant ammonia nitrogen removal coefficient are added together to obtain the comprehensive ammonia nitrogen removal coefficient, and the matrix total phosphorus removal coefficient and the plant total phosphorus removal coefficient are added together to obtain the comprehensive total phosphorus removal coefficient, with the upper limit of each comprehensive removal coefficient being 0.90.

[0016] Furthermore, on-site corrections are performed based on the comprehensive removal coefficients for ammonia nitrogen and total phosphorus: for subsidence water units with dissolved oxygen concentrations below 2 mg / L, the comprehensive removal coefficient for ammonia nitrogen is multiplied by 0.60 to obtain the corrected comprehensive removal coefficient for ammonia nitrogen; for subsidence water units with an average water depth exceeding 1.5 m, the plant ammonia nitrogen removal coefficient and the plant total phosphorus removal coefficient are each multiplied by 0.80 and then recalculated in the comprehensive removal coefficient calculation; the two corrections are performed cumulatively.

[0017] Furthermore, for adjacent subsidence aquifers with a surface elevation difference greater than 0.5m and connected surface runoff directions, the effluent ammonia nitrogen concentration of the upstream subsidence aquifer is obtained by subtracting the product of the influent ammonia nitrogen concentration and the comprehensive ammonia nitrogen removal coefficient of the upstream subsidence aquifer from the influent ammonia nitrogen concentration of the upstream subsidence aquifer. The influent of the downstream subsidence aquifer includes both the effluent discharged from the upstream subsidence aquifer and the mine water directly received by the downstream subsidence aquifer from the mine water network. The effluent ammonia nitrogen concentration is calculated by adding the product of the effluent ammonia nitrogen concentration and the flow rate of the upstream subsidence aquifer to the mine water. The influent ammonia nitrogen concentration of the downstream subsidence water unit is obtained by dividing the sum of the product of the raw water ammonia nitrogen concentration and the self-distributed flow of the downstream subsidence water unit by the sum of the outflow flow of the upstream subsidence water unit and the self-distributed flow of the downstream subsidence water unit. The influent total phosphorus concentration of the downstream subsidence water unit is calculated in the same way. The outflow flow of any subsidence water unit is equal to the sum of its own distributed flow and the outflow flow from the upstream subsidence water unit. The outflow flow of the upstreamst subsidence water unit is equal to its own distributed flow. The series relationship is passed down step by step from the upstreamst subsidence water unit according to the surface runoff direction.

[0018] Furthermore, step 2 includes three optimization objectives: the first objective is to maximize the total reduction of ammonia nitrogen load in the system, which is obtained by multiplying the allocated flow rate of each subsidence water unit by the influent ammonia nitrogen concentration and then by the comprehensive ammonia nitrogen removal coefficient, and summing the results for all subsidence water units; the second objective is to maximize the total reduction of total phosphorus load in the system, which is obtained by multiplying the allocated flow rate of each subsidence water unit by the influent total phosphorus concentration and then by the comprehensive total phosphorus removal coefficient, and summing the results for all subsidence water units; the third objective is to maximize the number of predicted compliant units, which is obtained by subtracting the product of the influent ammonia nitrogen concentration and the comprehensive ammonia nitrogen removal coefficient from the influent ammonia nitrogen concentration of each subsidence water unit to obtain the predicted effluent ammonia nitrogen concentration, and by subtracting the product of the influent total phosphorus concentration and the comprehensive total phosphorus removal coefficient from the influent total phosphorus concentration to obtain the predicted effluent total phosphorus concentration. Subsidence water units with a predicted effluent ammonia nitrogen concentration not exceeding 1.5 mg / L and a predicted effluent total phosphorus concentration not exceeding 0.3 mg / L are counted as compliant units.

[0019] Furthermore, the initial population contains 100 frogs, which are equally distributed into 10 frog groups. Within each frog group, the frog with the largest sum of the first, second, and third objectives is defined as the locally optimal frog, and the frog with the smallest sum of the first, second, and third objectives is defined as the locally worst frog. For each decision variable of the locally worst frog, a random number between 0 and 1 is generated. The random number is multiplied by the difference between the locally optimal frog and the locally worst frog on the corresponding decision variable. The product is then added to the corresponding decision variable of the locally worst frog to generate a new candidate frog. After the update, if the sum of the mine water allocation ratios of all subsidence water accumulation units is not equal to 100%, the deviation is distributed proportionally according to the existing mine water allocation ratios of each subsidence water accumulation unit and rounded up by a step size of 5%.

[0020] Furthermore, if the candidate new frog Pareto dominates the local worst frog, then the candidate new frog replaces the local worst frog; if the candidate new frog does not Pareto dominate the local worst frog, then one frog is randomly selected from the Pareto front set to replace the local best frog and the update operation is repeated; if the frog generated after replacement still does not Pareto dominate the local worst frog, then one frog that satisfies the total constraint that the sum of the mine water distribution ratios of all subsidence water accumulation units equals 100% is randomly regenerated to replace the local worst frog; the iteration terminates when the number of frogs in the Pareto front set and the three optimization objective values ​​of each frog no longer change after 20 consecutive iterations; the decision vector corresponding to the frog with the largest third objective is selected from the Pareto front set as the objective configuration scheme; if multiple frogs have the same third objective, the decision vector corresponding to the frog with the largest sum of the first and second objectives is selected as the objective configuration scheme.

[0021] Furthermore, in step 3, the dissolved oxygen concentration, ammonia nitrogen concentration, and total phosphorus concentration of each subsidence water unit are measured once every 14 days. When the ammonia nitrogen concentration of any subsidence water unit is higher than the predicted effluent ammonia nitrogen concentration of the corresponding subsidence water unit in step 2 for two consecutive measurements, or the total phosphorus concentration is higher than the predicted effluent total phosphorus concentration of the corresponding subsidence water unit in step 2 for two consecutive measurements, the most recently measured dissolved oxygen concentration, ammonia nitrogen concentration, and total phosphorus concentration are updated to the field information in step 1, and the process is returned to step 2 to re-execute the hybrid frog-jumping algorithm and Pareto multi-objective optimization coupling process to generate an updated target configuration scheme, until the ammonia nitrogen concentration and total phosphorus concentration of all subsidence water units meet the Class IV water quality requirements of the surface water environmental quality standard for 6 consecutive months.

[0022] By adopting the above technical solutions, the present invention has produced the following beneficial effects: The present invention establishes a series water quality transfer calculation model based on the difference in surface elevation between subsidence water accumulation units, which can accurately describe the concentration superposition effect of water from the upstream unit entering the downstream unit through surface runoff. It overcomes the defect of the prior art that designs each water accumulation area as an independent unit and ignores the hydraulic connection between units, so that the wetland design scheme can truly reflect the influent water quality of each unit under actual hydrological conditions, thereby improving the prediction accuracy and engineering reliability of the design scheme.

[0023] This invention uses the total reduction of ammonia nitrogen load, the total reduction of total phosphorus load, and the predicted number of compliant units as optimization objectives. It uses the Pareto ranking mechanism to search for the set of non-dominated optimal solutions in the objective space, avoiding the subjective problem of manually setting weight coefficients in traditional single-objective optimization or weighted summation methods. It can achieve global equilibrium among multiple mutually constraining water quality indicators, and ensure that as many water accumulation units as possible meet the surface water environmental quality standards by prioritizing the scheme with the largest number of compliant units.

[0024] This invention introduces a comprehensive removal coefficient correction mechanism for two on-site environmental factors: dissolved oxygen concentration and water depth. When the dissolved oxygen in the water is below a set threshold, an attenuation correction is applied to the ammonia nitrogen removal coefficient. When the water depth exceeds a set threshold, an attenuation correction is applied to the plant removal coefficient. The two corrections can be performed cumulatively, so that the removal coefficient used in the theoretical calculation can accurately reflect the weakening effect of adverse on-site conditions on the purification effect, reduce the deviation between theoretical prediction and actual operating effect, and improve the adaptability of the optimization scheme to complex on-site conditions.

[0025] This invention fully utilizes a population evolution strategy that combines local search and global shuffling in the hybrid frog jumping algorithm. Within each group, a local fine search is achieved by directional jumping from the worst-performing frog to the best-performing frog. In the global shuffling stage, the closure of the local search is broken by reordering and grouping, which effectively balances the algorithm's development and exploration capabilities, reduces the risk of getting trapped in local optima, and improves the efficiency and diversity of searching for Pareto front sets in high-dimensional decision spaces. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the longitudinal section structure and pollutant removal path of the constructed wetland treatment unit provided in an embodiment of the present invention.

[0027] Figure 2 This invention provides the principle of water quality transfer and concentration decay in a series-connected subsidence water accumulation unit.

[0028] Figure 3 The iterative convergence curve of the Pareto multi-objective optimization of the hybrid frog-leaping algorithm provided in this embodiment of the invention;

[0029] Figure 4 The distribution diagram of the Pareto front set in the two-dimensional target space formed by the first target and the second target, provided for embodiments of the present invention. Detailed Implementation

[0030] A method for the coordinated restoration of mine water ecological replenishment wetlands in subsidence and waterlogged areas includes the following steps:

[0031] Step 1: Divide the coal mining subsidence and water accumulation area into several subsidence and water accumulation units according to the surface elevation boundary, collect the water quality and hydrological parameters of each subsidence and water accumulation unit, record the relationship between the surface elevation difference and surface runoff direction between adjacent subsidence and water accumulation units, and record the total flow rate of mine water outflow.

[0032] Step 2: For each subsidence water accumulation unit, determine the decision variables including substrate type, planted vegetation, inlet location, and mine water allocation ratio. Concatenate the decision variables of all subsidence water accumulation units into a decision vector and define it as one frog. Randomly generate several frogs to form an initial population. The sum of the mine water allocation ratios of all subsidence water accumulation units in each frog is equal to 100%. Based on the decision vector of each frog and the field information collected in Step 1, calculate the comprehensive removal coefficient for each unit and calculate multiple optimization target values ​​corresponding to each frog accordingly. Perform Pareto sorting on all frogs in the population and distribute them to several frog groups at equal intervals. Within each frog group, update the decision vector of the worst frog in the local group based on the best frog in the local group and determine whether to replace it based on the Pareto dominance relationship. After completing the local search of all frog groups, mix all frogs, re-execute Pareto sorting and grouping, and enter the next iteration. After the iteration terminates, select the target configuration scheme from the Pareto front set.

[0033] Step 3: Construct artificial wetlands in each subsidence and water accumulation unit according to the target configuration scheme, based on the determined substrate type, plant species, and inlet location. Continuously pump mine water into each subsidence and water accumulation unit according to the determined mine water distribution ratio. After purification by the artificial wetlands, the mine water is replenished into the water body of the subsidence and water accumulation unit, thus completing the ecological replenishment wetland collaborative restoration of mine water in the subsidence and water accumulation area.

[0034] The implementation process of steps 1 and 2 will be explained in detail below using a specific engineering application scenario.

[0035] Located in the North Anhui Plain, a mining area has developed a coal mining subsidence and waterlogging zone covering approximately 3.6 square kilometers due to years of underground mining. The water depth ranges from 0.3m to 2.8m, with excessive levels of ammonia nitrogen and total phosphorus. The mine discharges approximately 12,000 cubic meters of mine water daily; direct discharge would exacerbate eutrophication of surrounding water bodies. To achieve resource utilization of the mine water and restore the aquatic ecological function of the subsidence area, this invention introduces the mine water into the subsidence and waterlogging area as an ecological replenishment source. Artificial wetlands are constructed in each waterlogged area to purify the water, forming an integrated and synergistic system of replenishment and restoration.

[0036] refer to Figure 1 , Figure 1 To illustrate a comprehensive diagram of various substrates and plant removal pathways, a single subsidence water unit is used as an example to demonstrate the longitudinal cross-sectional structure of the wetland treatment unit along the water flow direction and the migration and transformation pathways of pollutants in each media layer. To fully present the filling methods of the three substrate types and three plant species, and their corresponding pollutant removal mechanisms, Figure 1 The same cross-section sequentially shows three sections: zeolite matrix, biochar matrix, and gravel matrix, along with the corresponding three plant species: reeds, cattails, and calamus. In actual engineering implementation, based on the optimization results of the hybrid frog-jumping algorithm in step 2, each subsidence and water accumulation unit selects only one matrix type and one plant species for construction. A bottom sediment layer is set at the bottom of the cross-section, and a layer with a thickness of [missing information] is laid on top of the bottom sediment layer. A graded crushed stone cushion layer of m is used. The function of the crushed stone cushion layer is to support the upper matrix and prevent matrix particles from migrating downwards. Above the crushed stone cushion layer is the matrix layer, with a total thickness of m. The diagram, shown in m, is divided into three sections along the water flow direction for illustrative purposes. These sections are filled with zeolite, biochar, and gravel substrates respectively to demonstrate the filling effect of each substrate. The sections are separated by vertical dashed lines. The zeolite substrate filling thickness is [missing information]. m, the filling thickness of the biochar matrix is m, the filling thickness of the gravel matrix is m. The area above the top surface of the matrix layer is the water body region, and the water surface line is located approximately above the top surface of the matrix layer. At point m, three types of wetland plants are schematically distributed within the substrate layer and above the water surface to demonstrate their growth morphology and removal mechanisms. Reeds are planted above the zeolite section, cattails above the biochar section, and sweet flag above the gravel section. The stems of each plant grow upwards from the substrate surface, while their roots penetrate the substrate layer and extend downwards. A mine water inlet is located on the left side of the profile. concentration mg / L, total phosphorus concentration mg / L enters the treatment unit; a purified water outlet is located on the right side of the cross-section, and the effluent... Concentration dropped to mg / L, total phosphorus concentration decreased to mg / L. Arrows indicating the direction of water flow are marked along the flow path in the water body. The diagram also indicates three main pollutant removal pathways: the first pathway is... Ion exchange adsorption indicates that ammonium nitrogen is fixed through cation exchange in a zeolite matrix; the second pathway is... Complexation fixation indicates that phosphate ions undergo complexation precipitation through metal oxides on the surface of the biochar matrix; the third pathway is the nitrification-denitrification process of rhizosphere microorganisms, in which microorganisms around plant roots sequentially oxidize ammonium nitrogen to nitrate nitrogen and ultimately reduce it to nitrogen. It escapes to the water surface. At the bottom of the profile, there is also a gradient bar showing the change in pollutant concentration along the direction of water flow, indicating that the pollutant concentration gradually decreases from a high concentration at the inlet to a low concentration at the outlet.

[0037] The implementation process of step 1 is as follows.

[0038] A digital elevation model with a resolution of 0.5m was acquired using a drone equipped with lidar. Based on this model, the coal mining subsidence and water accumulation area was divided into several subsidence and water accumulation units according to the surface elevation boundaries. The surface elevation boundary refers to the closed or semi-closed catchment boundary formed by naturally formed watersheds or artificial embankments in the digital elevation model. Each subsidence and water accumulation unit corresponds to a relatively independent catchment depression. The criteria for division are: the surface elevation variation within a single subsidence and water accumulation unit should not exceed 1.0m to ensure relatively uniform hydraulic conditions; and there must be an identifiable elevation difference between adjacent subsidence and water accumulation units to provide a basis for subsequently determining the surface runoff direction. In this embodiment, the coal mining subsidence and water accumulation area was divided into 6 subsidence and water accumulation units, denoted as Unit A to Unit F. In optional embodiments, the number of units can be flexibly determined based on the actual area of ​​the subsidence area and the degree of topographic relief; larger areas with more complex topography should be divided into more subsidence and water accumulation units.

[0039] After the subsidence area was divided, water quality and hydrological parameters were collected for each subsidence water unit. Water quality parameters included dissolved oxygen concentration, ammonia nitrogen concentration, and total phosphorus concentration. Dissolved oxygen concentration directly determines the nitrification activity of aerobic microorganisms; once dissolved oxygen in the water is insufficient, the biological oxidation rate of ammonia nitrogen decreases significantly. This parameter plays a crucial role in the subsequent correction of the removal coefficient. Ammonia nitrogen concentration and total phosphorus concentration are core indicators characterizing the eutrophication level of the water body and are also direct inputs for calculating the three optimization objectives. Hydrological parameters included the water accumulation area and average water depth. The water accumulation area limits the maximum area where constructed wetlands can be deployed, while the average water depth restricts the effective contact ratio between emergent plant roots and the water body. When the water is too deep, the plant rhizosphere can only cover the upper layer of the water, reducing the interception and absorption efficiency of pollutants in the middle and lower layers.

[0040] In this embodiment, the measured data for the six subsidence and water accumulation units are as follows: Unit A: Water accumulation area 0.8 square kilometers, average water depth 0.6 m, dissolved oxygen concentration 3.5 mg / L, ammonia nitrogen concentration 8.2 mg / L, total phosphorus concentration 1.6 mg / L. Unit B: Water accumulation area 0.5 square kilometers, average water depth 1.2 m, dissolved oxygen concentration 2.8 mg / L, ammonia nitrogen concentration 6.5 mg / L, total phosphorus concentration 1.2 mg / L. Unit C: Water accumulation area 0.7 square kilometers, average water depth 1.8 m, dissolved oxygen concentration 1.5 mg / L, ammonia nitrogen concentration 9.0 mg / L, total phosphorus concentration 2.0 mg / L. Unit D: Water accumulation area 0.4 square kilometers, average water depth 0.5 m, dissolved oxygen concentration 4.2 mg / L, ammonia nitrogen concentration 5.8 mg / L, total phosphorus concentration 0.9 mg / L. Unit E has a water area of ​​0.6 square kilometers, an average water depth of 1.0 m, a dissolved oxygen concentration of 3.0 mg / L, an ammonia nitrogen concentration of 7.0 mg / L, and a total phosphorus concentration of 1.4 mg / L. Unit F has a water area of ​​0.6 square kilometers, an average water depth of 2.2 m, a dissolved oxygen concentration of 1.8 mg / L, an ammonia nitrogen concentration of 10.5 mg / L, and a total phosphorus concentration of 2.5 mg / L.

[0041] Simultaneously, the relationship between the surface elevation difference and surface runoff direction between adjacent subsidence water accumulation units is recorded. The surface elevation difference is obtained from the average elevation difference at the boundaries of adjacent subsidence water accumulation units in the digital elevation model. The surface runoff direction is determined according to the elevation difference direction, with surface runoff flowing from subsidence water accumulation units with higher elevations to those with lower elevations. In this embodiment, the specific situations are as follows: Unit A and Unit C are adjacent, with a surface elevation difference of 0.8m, and the surface runoff direction is from Unit A to Unit C; Unit C and Unit E are adjacent, with a surface elevation difference of 0.6m, and the surface runoff direction is from Unit C to Unit E; Unit D and Unit F are adjacent, with a surface elevation difference of 0.7m, and the surface runoff direction is from Unit D to Unit F; Unit B and Unit D are adjacent, but the surface elevation difference is only 0.3m, which does not reach the 0.5m threshold required for subsequent series transmission, so the two are considered independent in the water quality transmission calculation. In the remaining unlisted adjacent relationships, the difference in surface elevation does not exceed 0.5m.

[0042] In addition, the total outflow of mine water is recorded. This flow rate represents the total volume of mine water pumped from underground to the surface daily under normal production conditions; in this embodiment, it is 12,000 cubic meters per day. After preliminary sedimentation on the surface to remove suspended solids, the mine water is distributed to various subsidence and water accumulation units through a distribution network. All the information collected and recorded above is collectively referred to as field information and is used in step 2.

[0043] refer to Figure 2 , Figure 2It consists of two parts. The upper part is a schematic diagram of the topographic longitudinal profile along the series path. The diagram shows the surface undulation outline of the coal mining subsidence area, with three subsidence and water accumulation units distributed horizontally, labeled as Unit A, Unit C, and Unit E. The surface elevation of Unit A is [missing information]. m, the surface elevation of unit C is m, the surface elevation of unit E is m. The water accumulation area of ​​each unit is represented by water filling the area, with the water surface line located within its respective depression. Since the elevation difference between unit A and unit C is... m, greater than the threshold for serial transmission m, therefore, there is a surface runoff transfer relationship between the two, and the runoff direction is indicated by an arc arrow in the figure. Similarly, the elevation difference between unit C and unit E is m. m, also satisfying the series condition, and the runoff transfer between the two is marked with an arc-shaped arrow. The mine water distribution ratio is marked above each unit; unit A receives... Mine water, received by Unit C Unit E receives The right side of the map has an elevation scale, marked with... m、 m and Three elevation values ​​(m). The corresponding values ​​are also labeled above each unit. Concentration and overall removal coefficient Value, water inlet of unit A concentration mg / L, comprehensive removal coefficient Unit C receives water transferred from upstream. Concentration dropped to mg / L, comprehensive removal coefficient after on-site correction The lower half is The graph shows the gradual decay curves of total phosphorus concentration along a series path. The horizontal axis, from left to right, is labeled with four nodes: raw mine water, effluent from unit A, effluent from unit C, and effluent from unit E. The vertical axis represents concentration in mg / L. Two decay curves are plotted in the graph, one of which is... Concentration change curve, concentrations at each node are as follows: , , and mg / L; the other is the total phosphorus concentration change curve, with the concentrations at each node being mg / L. , , and mg / L. The corresponding concentration value is labeled next to each data point. The removal rate is labeled between adjacent nodes; for example, the removal rate after treatment of raw mine water in unit A. Removal rate reached The figure also shows two horizontal dashed lines corresponding to the Class IV water quality limits of the surface water environmental quality standards, namely... concentration mg / L and total phosphorus concentration mg / L.

[0044] The implementation process of step 2 is as follows.

[0045] The overall idea of ​​step 2 is to encode the wetland design parameters of each subsidence and water accumulation unit into a unified scheme, and then use a hybrid frog-jumping algorithm to conduct a population search in a discrete-continuous mixed scheme space. Simultaneously, a Pareto multi-objective optimization framework is introduced to handle the three competing optimization objectives. The hybrid frog-jumping algorithm is used instead of traditional genetic algorithms or particle swarm optimization algorithms primarily for two reasons: First, the hybrid frog-jumping algorithm divides the population into multiple frog groups for local searches, and then periodically merges and rearranges them. This "divide and conquer—convergence" two-layer structure achieves a balance between local fine-grained search and global diversity maintenance. Second, within each frog group, only the worst individual is updated in each round, resulting in a computational cost far lower than the crossover and mutation operations performed on the entire population in genetic algorithms. Furthermore, the introduction of Pareto dominance eliminates the need to merge multiple objectives into a single scalar during the sorting and replacement stages, preserving the trade-off information between different objectives and avoiding subjective bias caused by manually setting objective weights.

[0046] First, the decision variables are defined. Four types of decision variables are determined for each subsidence and water accumulation unit.

[0047] Category 1 is the substrate type, selecting one from zeolite, gravel, and biochar. Zeolite's layered aluminosilicate framework endows it with excellent cation exchange properties, exhibiting selective adsorption capacity for ammonium nitrogen in water; gravel has good hydraulic conductivity and is inexpensive as a traditional wetland filler, but its active adsorption effect on pollutants is limited; biochar's surface is rich in oxygen-containing functional groups and microporous structures, enabling it to both chemically adsorb and fix phosphates and provide numerous attachment sites for microorganisms. Category 2 is the plant species, selecting one from reeds, cattails, and calamus. Reeds have deep, well-developed root systems and large aboveground biomass, exhibiting outstanding rhizosphere absorption capacity for ammonia nitrogen; cattails demonstrate highly efficient absorption of total phosphorus in eutrophic waters; calamus is relatively tolerant to water quality fluctuations and low temperatures, but its absorption capacity per unit area for ammonia nitrogen and total phosphorus is slightly lower than the former two. Category 3 is the inlet location, selecting one from one of the following: one at one end of the long side, one in the middle of the long side, or one dispersed at the four corners. One type of water inlet is positioned along the long side of the unit to form a push-flow hydraulic path. As the mine water flows from one end to the other, it makes full contact with the substrate and plant roots, resulting in a longer hydraulic residence time. This is suitable for subsidence water accumulation units with moderate area and a narrow shape. Another type is positioned along the middle of the long side, allowing the mine water to diffuse to both sides. This is suitable for larger units with a wider area. A third type is positioned at the four corners, allowing water to be injected simultaneously from four directions, minimizing hydraulic short-circuiting and dead zones. This is suitable for units with near-square or irregular shapes. The fourth type is the mine water allocation ratio, which is the percentage of the total mine water outflow allocated to the corresponding subsidence water accumulation unit, with a step size of 5% and a range of 5% to 60%. The lower limit of 5% ensures that each subsidence water accumulation unit receives the minimum replenishment required to maintain basic wetland operation, while the upper limit of 60% prevents excessive concentration of mine water in a single subsidence water accumulation unit, which could lead to hydraulic overload. In an optional implementation, the step size can be adjusted to 10% to accelerate the search, or to 2% to obtain a more refined allocation scheme; the upper and lower limits of the range can also be appropriately widened or narrowed according to the differences in area and bearing capacity of each subsidence water accumulation unit.

[0048] The decision variables of all subsidence water accumulation units are concatenated into a single decision vector. This embodiment has six subsidence water accumulation units, each containing four types of decision variables, thus the single decision vector contains 24 components. Discrete variables are encoded as integers: zeolite is 1 for matrix type, gravel is 2, and biochar is 3; reeds are 1, cattails are 2, and sweet flag is 3; the location of the inlet is 1 at one end of the long side, 2 at the middle of the long side, and 3 at the four corners. The mine water allocation ratio is expressed as a percentage integer, such as 15 representing 15%. Each decision vector is defined as a frog, and each frog corresponds to one overall wetland configuration scheme encompassing all subsidence water accumulation units.

[0049] A random number of frogs are generated to form an initial population; in this embodiment, this is 100 frogs. The specific generation method is as follows: For each frog in each subsidence water accumulation unit, one substrate type is randomly selected with equal probability from three options; one plant type is randomly selected with equal probability from three options; and one inlet location is randomly selected with equal probability from three options. Regarding the mine water allocation ratio, each subsidence water accumulation unit is first randomly assigned an initial value between 5 and 60, and which is an integer multiple of 5. Then, it is checked whether the sum of the mine water allocation ratios of all subsidence water accumulation units is exactly equal to 100%. If it is not equal to 100%, the deviation is calculated. This deviation is then distributed to each subsidence water accumulation unit according to the proportion of its current mine water allocation ratio to the total ratio. After distribution, each value is rounded up in 5% increments. This process is repeated until the sum of the mine water allocation ratios is exactly equal to 100%. This constraint ensures that the total outflow of mine water is completely allocated without omission or over-allocation.

[0050] Next, based on the decision vector of each frog and the on-site information collected in step 1, the comprehensive removal coefficient is calculated unit by unit.

[0051] For each subsidence unit of each frog in the population, removal coefficients were determined based on the substrate type and plant species in the decision vector. The rules for determining the substrate ammonia nitrogen removal coefficient were: 0.60 for zeolite, 0.50 for biochar, and 0.30 for gravel. The rules for determining the substrate total phosphorus removal coefficient were: 0.60 for biochar, 0.40 for zeolite, and 0.20 for gravel. The rules for determining the plant ammonia nitrogen removal coefficient were: 0.30 for reeds, 0.20 for cattails, and 0.15 for sweet flag. The rules for determining the plant total phosphorus removal coefficient were: 0.30 for cattails, 0.20 for reeds, and 0.15 for sweet flag. These values ​​are derived from published pilot-scale experimental reports of constructed wetlands and operational statistics from multiple engineering cases. The use of lookup tables allows for rapid evaluation of the scheme without relying on complex hydrodynamic models.

[0052] The comprehensive ammonia nitrogen removal coefficient is obtained by adding the matrix ammonia nitrogen removal coefficient and the plant ammonia nitrogen removal coefficient, and the comprehensive total phosphorus removal coefficient is obtained by adding the matrix total phosphorus removal coefficient and the plant total phosphorus removal coefficient. Addition, rather than multiplication, is used because matrix adsorption and plant absorption in constructed wetlands remove pollutants through two independent pathways: physicochemical and biological processes. Within the range where removal efficiency does not approach saturation, the contributions of the two can be approximately superimposed. The upper limit for each comprehensive removal coefficient is set at 0.90; that is, when the sum of the two coefficients exceeds 0.90, 0.90 is taken. No single treatment unit can achieve 100% removal in actual operation. Constrained by engineering factors such as matrix adsorption saturation, hydraulic short-circuiting, and effluent residue, 0.90 is a reasonable engineering upper limit for the removal efficiency of a single unit.

[0053] Let's demonstrate a complete lookup and summation process using unit D as an example. Suppose a frog selects gravel as the substrate type and calamus as the plant in unit D. The ammonia nitrogen removal coefficient for the substrate is 0.30, and the ammonia nitrogen removal coefficient for the plant is 0.15. The sum of the two is 0.45, which does not exceed the upper limit, so the overall ammonia nitrogen removal coefficient is 0.45. The total phosphorus removal coefficient for the substrate is 0.20, and the total phosphorus removal coefficient for the plant is 0.15. The sum of the two is 0.35, so the overall total phosphorus removal coefficient is 0.35.

[0054] After obtaining the initial comprehensive removal coefficient, two on-site corrections are required. The first correction targets subsidence water units with dissolved oxygen concentrations below 2 mg / L. When dissolved oxygen in the water drops below 2 mg / L, the aerobic nitrifying bacteria community on which ammonia nitrogen nitrification depends can hardly maintain normal metabolism, and the nitrification rate drops sharply. To reflect this constraint, the comprehensive ammonia nitrogen removal coefficient is multiplied by 0.60 to obtain the corrected comprehensive ammonia nitrogen removal coefficient. In this embodiment, the dissolved oxygen concentration in unit C is 1.5 mg / L and the dissolved oxygen concentration in unit F is 1.8 mg / L, both of which are below 2 mg / L, requiring this correction. The second correction targets subsidence water units with an average water depth exceeding 1.5 m. Once the water depth exceeds 1.5 m, the proportion of the root distribution range of common emergent plants such as reeds, cattails, and calamus relative to the total water depth decreases significantly, and the interception and absorption efficiency of ammonia nitrogen and total phosphorus in the water by the plant rhizosphere decreases accordingly. The specific method involves multiplying the plant ammonia nitrogen removal coefficient and the plant total phosphorus removal coefficient by 0.80 each, and then adding the corrected plant removal coefficient back to the substrate removal coefficient to obtain the corrected comprehensive removal coefficient. In this embodiment, the average water accumulation depth of unit C is 1.8m and the average water accumulation depth of unit F is 2.2m, both exceeding 1.5m, thus requiring this correction. The two corrections can be applied cumulatively; that is, if a subsidence water accumulation unit simultaneously meets the conditions of dissolved oxygen concentration below 2mg / L and average water accumulation depth exceeding 1.5m, both corrections must be applied.

[0055] The following is a complete calculation of the superposition correction using unit F as an example. Assume that unit F uses biochar as the substrate and cattail as the plant. The substrate ammonia nitrogen removal coefficient is 0.50, and the plant ammonia nitrogen removal coefficient is 0.20. Since the average water depth of 2.2m exceeds 1.5m, the plant ammonia nitrogen removal coefficient of 0.20 is multiplied by 0.80 to obtain 0.16, and the plant total phosphorus removal coefficient of 0.30 is multiplied by 0.80 to obtain 0.24. The substrate ammonia nitrogen removal coefficient of 0.50 is added to the depth-corrected plant ammonia nitrogen removal coefficient of 0.16, resulting in 0.66. Since the dissolved oxygen concentration of 1.8mg / L is lower than 2mg / L, 0.66 is multiplied by 0.60 to obtain 0.396, which is the corrected comprehensive ammonia nitrogen removal coefficient for unit F. The substrate total phosphorus removal coefficient of 0.60 is added to the depth-corrected plant total phosphorus removal coefficient of 0.24, resulting in 0.84. Since dissolved oxygen correction does not involve total phosphorus, the comprehensive total phosphorus removal coefficient for unit F is 0.84.

[0056] Next, we address the cascading water quality transfer between adjacent subsidence water accumulation units. When the surface elevation difference is greater than 0.5m and the surface runoff flows in a continuous direction, the effluent from the upstream subsidence water accumulation unit will flow into the downstream subsidence water accumulation unit along the surface runoff, directly impacting the downstream influent water quality due to the upstream purification effect. This mechanism is significant at the optimization level. If the optimization algorithm configures a substrate-plant combination with strong removal capacity in the upstream unit, the actual influent concentration received by the downstream unit will be significantly lower than its original measured concentration. The downstream unit can then adopt a lower-cost configuration or one that focuses on other objectives, thus achieving better resource allocation at the system-wide level. If this cascading transfer is not included in the calculation, each subsidence water accumulation unit is treated as an isolated entity during objective evaluation. The algorithm will fail to capture this upstream-downstream coupling effect, and the output solution will often deviate from the optimal solution in actual operation.

[0057] The specific calculation method for series transfer is as follows: First, the effluent ammonia nitrogen concentration of the upstream subsidence water unit is obtained by subtracting the product of the influent ammonia nitrogen concentration and the comprehensive ammonia nitrogen removal coefficient of the upstream subsidence water unit from the influent ammonia nitrogen concentration of the upstream subsidence water unit. Since the influent of the downstream subsidence water unit includes both the effluent discharged from the upstream subsidence water unit and the mine water directly received by the downstream subsidence water unit from the mine water pipeline network, the influent ammonia nitrogen concentration of the downstream subsidence water unit is obtained by dividing the sum of the product of the effluent ammonia nitrogen concentration and the flow rate of the upstream subsidence water unit plus the product of the raw mine water ammonia nitrogen concentration and the self-distributed flow rate of the downstream subsidence water unit by the sum of the flow rate of the upstream subsidence water unit and the self-distributed flow rate of the downstream subsidence water unit. The transfer of total phosphorus follows the same pattern: the total phosphorus concentration in the effluent of the upstream subsidence aquifer is calculated by multiplying the total phosphorus concentration of the raw mine water by the total phosphorus concentration of the downstream subsidence aquifer itself, and then dividing the sum of the effluent flow rate of the upstream and downstream subsidence aquifers by the sum of their respective flow rates. The effluent flow rate of any subsidence aquifer is equal to the sum of its own flow rate and the flow rate from the next upstream subsidence aquifer. The effluent flow rate of the most upstream subsidence aquifer is equal to its own flow rate. The series connection is calculated sequentially downwards from the most upstream subsidence aquifer, following the direction of surface runoff. In multi-stage series connections, the calculation must begin from the most upstream subsidence aquifer and proceed sequentially downwards.

[0058] Taking the series connection of Unit A—Unit C—Unit E as an example. The influent ammonia nitrogen concentration of Unit A is 8.2 mg / L, collected in step 1. If the comprehensive ammonia nitrogen removal coefficient of Unit A is 0.90, then the influent ammonia nitrogen concentration transferred to Unit C is... mg / L, of which The influent ammonia nitrogen concentration for Unit A is... The comprehensive ammonia nitrogen removal coefficient for Unit A is given. Unit A is the upstream subsidence water accumulation unit, and its flow rate is equal to its own allocated flow rate. Unit C receives 0.82 mg / L effluent from Unit A, and also directly receives raw mine water from the mine water network. The influent ammonia nitrogen concentration of Unit C is calculated using a flow-weighted mixing method: the product of the effluent ammonia nitrogen concentration of Unit A (0.82 mg / L) and the flow rate of Unit A, plus the product of the raw mine water ammonia nitrogen concentration of 8.2 mg / L and the allocated flow rate of Unit C, is divided by the sum of the flow rate of Unit A and the allocated flow rate of Unit C. The total phosphorus concentration is calculated in the same way. After completing the purification calculation for Unit C, the effluent concentration of Unit C is then mixed with the mine water directly received by Unit E using the same flow-weighted mixing method and transferred to Unit E.

[0059] refer to Figure 3 , Figure 3 The graph consists of three subgraphs arranged from top to bottom, sharing a horizontal axis to represent the iteration rounds, with a range of [missing information]. to The first subplot shows the change in the total ammonia nitrogen load reduction of the first objective, i.e., with the vertical axis in kg / d. Two curves are plotted in the figure: one is the current population optimum curve, representing the maximum value of the Pareto front set in the first objective dimension in each iteration; this curve fluctuates significantly in the early stages and tends to stabilize in the later stages. The other is the historical optimum curve, representing the cumulative maximum value of the first objective up to the current iteration; this curve monotonically increases and eventually converges to approximately [missing value]. kg / d, with the final convergence value marked by a horizontal dashed line in the figure. In the iteration rounds... to The interval within which the convergence interval is marked indicates that the Pareto front set is in a continuous state. No changes occurred during the iterations, satisfying the termination condition. The second sub-graph illustrates the iterative process of the second objective, namely the total reduction of the system's total phosphorus load, with the vertical axis also in kg / d. The structure of the graph is the same as the first sub-graph. Figure 1 The current population optimum curve and the historical optimum curve reflect the search, exploration, and cumulative improvement of the second objective during population evolution, respectively. The historical optimum eventually converges to approximately [value missing]. kg / d. The third subplot is a dual-axis plot. The left axis represents the third objective, i.e., the number of predicted target units, in units, with values ​​ranging from [number]. to The right vertical axis represents the size of the Pareto front set, in units of individuals (i.e., the number of non-dominant frog individuals in the Pareto front set). The third target curve is stepped, reflecting the number of achieving units as the iteration progresses from the initial value... The number gradually increased to indivual, One, finally in about the 1st Rear stability reached The final convergence value is marked with a horizontal dashed line. The Pareto front set size curve reflects the change in the number of non-dominant frog individuals, from the initial... Only gradually increase to a stable level The presence of "only" indicates that the population continuously discovers new Pareto optimal solutions during the iteration process until the frontier set tends to saturate. All three subgraphs are marked with convergence intervals at the end of the iteration, indicating that all three objectives simultaneously satisfy the termination condition.

[0060] After determining the overall removal coefficient and influent concentration for all subsidence water units, multiple optimization target values ​​are calculated for each frog. These multiple optimization target values ​​include three optimization objectives.

[0061] The primary objective is to maximize the total reduction in ammonia nitrogen load across the system. For each subsidence area, the contribution to ammonia nitrogen reduction is calculated by multiplying the allocated flow rate by the influent ammonia nitrogen concentration and then by the comprehensive ammonia nitrogen removal coefficient. The allocated flow rate equals the total outflow of mine water multiplied by the mine water allocation ratio. The sum of the ammonia nitrogen reduction contributions for all subsidence areas is the primary objective. Pursuing this primary objective implies a preference for allocating more mine water to subsidence areas with higher comprehensive ammonia nitrogen removal coefficients.

[0062] The second objective is to maximize the total phosphorus load reduction of the system. The calculation method is symmetrical to the first objective, simply replacing ammonia nitrogen concentration and the comprehensive ammonia nitrogen removal coefficient with total phosphorus concentration and the comprehensive total phosphorus removal coefficient, respectively. It is worth noting that the first and second objectives often compete: the substrate-plant combination most favorable for ammonia nitrogen removal (e.g., zeolite combined with reeds, achieving a comprehensive ammonia nitrogen removal coefficient of 0.90) performs far worse in terms of total phosphorus (a comprehensive total phosphorus removal coefficient of only 0.60) than the combination of biochar and cattail (which achieves a comprehensive total phosphorus removal coefficient of 0.90). It is precisely this competitive relationship between objectives that makes the Pareto multi-objective optimization framework necessary. It eliminates the need to manually set fixed weights between the two objectives, instead preserving a set of non-dominated solutions that achieve different balances among the different objectives for decision-makers to choose from.

[0063] The third objective is to maximize the number of predicted compliant units. In the calculation, for each subsidence area, the predicted effluent ammonia nitrogen concentration is obtained by subtracting the product of the influent ammonia nitrogen concentration and the comprehensive ammonia nitrogen removal coefficient from the influent ammonia nitrogen concentration. Similarly, the predicted effluent total phosphorus concentration is obtained by subtracting the product of the influent total phosphorus concentration and the comprehensive total phosphorus removal coefficient from the influent total phosphorus concentration. When the predicted effluent ammonia nitrogen concentration does not exceed 1.5 mg / L and the predicted effluent total phosphorus concentration does not exceed 0.3 mg / L, the corresponding subsidence area is counted as a compliant unit. The total number of compliant units is the third objective. 1.5 mg / L and 0.3 mg / L correspond to the limits for ammonia nitrogen and total phosphorus in Class IV surface water quality standards, respectively. The third objective measures the coverage area of ​​compliance rather than the total reduction. Even if the first and second objective values ​​for a particular subsidence area are considerable, if the high reduction is concentrated in a few subsidence areas while the effluent from most other areas still exceeds the standard, the value of the third objective will still be very low. Incorporating the third objective into the optimization effectively prevents the algorithm from producing solutions that are "locally compliant but globally uneven," guiding the search towards a direction where all subsidence and water accumulation units generally meet the requirements.

[0064] This completes the evaluation of the target value for each frog in the population. The next step is to proceed with the iterative optimization process that couples the hybrid frog jumping algorithm with Pareto dominance.

[0065] First, perform Pareto sorting on all 100 frogs in the population. The sorting rule is: if frog A is no worse than frog B in the first, second, and third objectives, and is strictly better than frog B in at least one optimization objective, then frog A is said to Pareto dominate frog B. Frogs that are not Pareto dominated by any other frog in the population constitute the current Pareto front set. Pareto sorting can be implemented by a layer-by-layer peeling method: first, traverse all frogs and find all individuals that are not dominated by any other frog, marking them as the first front layer; temporarily remove the first front layer, and repeat the same operation on the remaining frogs to obtain the second front layer; peel layer by layer until all frogs are assigned to a front layer. The first front layer is the Pareto front set.

[0066] refer to Figure 4 The horizontal axis represents the primary objective, namely the total reduction in ammonia nitrogen load in the system, expressed in kg / d, with a range of values. to The vertical axis represents the second objective, namely the total reduction in the system's total phosphorus load, expressed in kg / d, with a range of values. to The diagram contains a large number of data points. Points scattered in the lower part of the target space represent dominated solutions, i.e., frog individuals that do not belong to the first front layer in the Pareto ranking. These solutions are superior to other solutions in at least one target dimension. Points distributed in the upper right region of the target space are... There are several Pareto front solutions, and these non-dominated frog individuals constitute the Pareto front set, connected by dashed lines to form the front boundary. Pareto front solutions are divided into two categories based on the value of the third objective: those predicting the number of units achieving the objective are... The solutions are represented by pentagrams, and the predicted number of qualified units is [number missing]. The solutions are represented by diamonds. Among all Pareto front solutions, according to the selection rules for the target configuration scheme, the solution with the largest value of the third objective is first selected. Then, when the third objectives are equal, the solution with the largest sum of the first and second objectives is selected. The final selected target configuration scheme is located at coordinate... The figure shows a magnified pentagram with a leader line indicating the location of the three targets. The total ammonia nitrogen load reduction corresponding to this scheme is... kg / d, the total reduction in the system's total phosphorus load is kg / d, the predicted number of compliant units is The shape of the Pareto front boundary indicates a certain trade-off between the first and second objectives. That is, increasing the total reduction of ammonia nitrogen in the front set may lead to a decrease in the total reduction of total phosphorus, and vice versa. This reflects the typical characteristics of the Pareto optimal solution set in a multi-objective optimization problem.

[0067] The 100 sorted frogs are arranged from best to worst according to their Pareto front level, with the first front level at the front, followed by the second, and so on. They are then evenly distributed into 10 frog groups, with 10 frogs in each group. The evenly distributed operation is as follows: the first-ranked frog goes to the first group, the second to the second, and so on, until the 10th frog returns to the first group, and so on, until all 100 frogs are distributed. This method ensures that each frog group contains both high-quality solutions from the Pareto front set and lower-quality solutions from deeper front levels. The former provides clear guidance for local search, while the latter provides room for improvement. If high-quality solutions are concentrated in a few frog groups, the differences within those groups will be small, and local search will hardly produce effective updates; while frog groups with low-quality solutions will lack high-quality guidance and will be difficult to improve efficiently. The evenly distributed strategy balances this contradiction.

[0068] Local searches are performed within each frog swarm. Taking swarm 1 as an example, among the 10 frogs in this swarm, the frog with the largest sum of the first, second, and third objectives is defined as the locally optimal frog, and the frog with the smallest sum of the first, second, and third objectives is defined as the locally worst frog. Here, the direct summation of the three objective values ​​is used as a scalar scale to determine the locally optimal and locally worst frogs; this is only used for role assignment within the frog swarm and does not replace the role of Pareto dominance in the replacement decision.

[0069] Update the decision vector of the locally worst frog. The update operation is performed on a per-decision-variable basis: for each decision variable of the locally worst frog, generate a uniformly random number between 0 and 1. , random number Multiply the difference between the locally optimal and locally worst frogs on their corresponding decision variables, and then add the product to the corresponding decision variable of the locally worst frog. Taking the mine water allocation ratio as an example, suppose the locally worst frog allocates 10% of the mine water in unit A, and the locally optimal frog allocates 25% of the mine water in unit A. (Random number...) The update quantity is After superposition, it becomes Rounding down to 15% with a 5% step size. For the three discrete decision variables—substrate type, plant species, and inlet location—the updated values ​​are rounded to integers and modulo 3 (the total number of options) to ensure the encoded values ​​fall within the valid range of 1 to 3. The essence of this update operation is to guide the locally worst frog towards the locally best frog by a random step size, using random numbers. The introduction of this feature makes the convergence magnitude different for each decision variable, thus avoiding the problem of all decision variables moving in the same proportion and causing the search path to become monotonous.

[0070] After all decision variables are updated, check the total constraint. If the sum of the mine water distribution ratios of all subsidence water accumulation units is not equal to 100%, calculate the deviation, distribute the deviation proportionally according to the existing mine water distribution ratio of each subsidence water accumulation unit, round it up in 5% increments, and repeatedly adjust until the total constraint is met, generating candidate new frogs.

[0071] The replacement is then determined by Pareto dominance. If the candidate frog Pareto dominates the local worst frog (i.e., at least one of the first, second, and third objectives is strictly better than the local worst frog and the rest are not worse), then the candidate frog replaces the local worst frog. If the candidate frog does not Pareto dominate the local worst frog, it means that using the locally optimal frog within the frog swarm as a guide has failed to achieve improvement. In this case, a frog is randomly selected from the Pareto front set to replace the locally optimal frog, and the above update operation is repeated. The significance of this step is that when the locally optimal frog within the frog swarm is in a position in the search space that is unfavorable to guiding the improvement of the local worst frog, the globally non-dominated solution in the Pareto front set may be located in a completely different search direction, and using it as a guide can open up new improvement paths. If the frog generated after replacement still does not Pareto dominate the local worst frog, it means that no effective improvement can be found for the current local worst frog, whether from within the frog population or from the global frontier. In this case, one frog is randomly generated to replace the local worst frog, satisfying the total constraint that the sum of the mine water distribution ratios of all subsidence water units equals 100%, in order to inject new randomness and maintain population diversity.

[0072] After the 10 frog groups have completed their local searches in turn, all 100 frogs are mixed together, and the Pareto sort is re-executed. This involves redetermining the Pareto front set and its levels, and then redistributing the frogs to the 10 groups at equal intervals for the next iteration. Each round of mixing and rearranging allows frogs to move between different groups. A frog that was the worst locally in a group in the previous round may be grouped with frogs from different search regions in the new round, thus gaining new local optimal guidance. This continuous flow of information within the population is the key difference between the hybrid frog-jumping algorithm and simple multi-population parallel search.

[0073] Regarding the iteration termination condition: the iteration terminates when, after 20 consecutive iterations, the number of frogs in the Pareto front set and the three objective values ​​of each frog no longer change. In other words, if the... At the end of the first iteration, the Pareto front set and the first iteration... If the rounds are identical, the number of frogs in the set is the same, and the first, second, and third objective values ​​of each frog are identical, then the Pareto front is considered to have converged, and further iterations will not produce new improvements. In an optional implementation, the termination condition can also be set to a fixed maximum number of iteration rounds (such as 200 or 500 rounds), or set to the maximum variation of the objective value of each frog in the Pareto front set within a certain number of consecutive rounds being lower than a predetermined convergence threshold.

[0074] After the iteration terminates, a target configuration scheme is selected from the Pareto front set. The selection rule is: the decision vector corresponding to the frog with the largest third objective is selected from the Pareto front set as the target configuration scheme. The reason for prioritizing the frog with the largest third objective is that in the ecological restoration of coal mining subsidence and water accumulation areas, the general compliance of each subsidence and water accumulation unit is more practically valuable than simply pursuing the total reduction of the system—regulatory departments usually require all discharge outlets to meet water quality standards, rather than just achieving total quantity compliance. If multiple frogs have the same third objective, the decision vector corresponding to the frog with the largest sum of the first and second objectives is selected as the target configuration scheme, that is, maximizing the total reduction under the premise that the number of compliant units is the same. In an optional implementation, the engineering decision-maker can adjust the selection rule according to the specific restoration priority, for example, in areas where total phosphorus exceedance is more severe, the frog with the largest second objective is prioritized.

[0075] Decoding the decision vector of the selected frog yields the matrix type, planted vegetation, inlet location, and mine water distribution ratio for each subsidence water accumulation unit, which constitutes the complete content of the target configuration scheme.

[0076] The following is a possible optimization result in this embodiment. After iterative optimization, the Pareto front set contains 12 frogs, and the three optimization target values ​​corresponding to the frog with the largest third objective are: the first objective is 62.4 kg per day, the second objective is 15.8 kg per day, and the third objective is 5 target units. The decision vector decoding results for the frog are as follows: Unit A—matrix type zeolite, planted with reeds, inlet located at one end of the long side, mine water allocation ratio 20%; Unit B—matrix type biochar, planted with cattails, inlet located in the middle of the long side, mine water allocation ratio 15%; Unit C—matrix type biochar, planted with reeds, inlet located at the four corners, mine water allocation ratio 25%; Unit D—matrix type gravel, planted with calamus, inlet located at one end of the long side, mine water allocation ratio 10%; Unit E—matrix type zeolite, planted with cattails, inlet located in the middle of the long side, mine water allocation ratio 15%; Unit F—matrix type biochar, planted with cattails, inlet located at the four corners, mine water allocation ratio 15%. The sum of the mine water allocation ratios for each subsidence water accumulation unit is 20% + 15% + 25% + 10% + 15% + 15%, which equals 100%, satisfying the total amount constraint.

[0077] Several noteworthy allocation characteristics can be identified from this scheme. Unit C received the highest mine water allocation ratio of 25%, which is consistent with its large area (0.7 square kilometers) and high pollution concentration (ammonia nitrogen 9.0 mg / L, total phosphorus 2.0 mg / L). Allocating more mine water means a greater hydraulic load to drive the substrate and plants to play a removal role. At the same time, Unit C used a combination of biochar and reeds with inlets distributed in four corners. The strong adsorption of total phosphorus by biochar compensated for the high total phosphorus inlet pressure faced by Unit C. The efficient absorption of ammonia nitrogen by reeds was further utilized after the upstream Unit A (zeolite-reed combination, comprehensive ammonia nitrogen removal coefficient 0.90) had significantly reduced the inlet ammonia nitrogen concentration transferred to Unit C. The four-corner distribution also reduced hydraulic dead zones due to the large area of ​​Unit C. Unit D received only 10% of the mine water allocation because its initial ammonia nitrogen and total phosphorus concentrations were the lowest among the six units (ammonia nitrogen 5.8 mg / L, total phosphorus 0.9 mg / L), resulting in the lowest treatment pressure. Combined with a low-cost gravel-calamus combination, it met the compliance requirements without consuming excessive mine water resources. The algorithm automatically discovered this "upstream enhanced purification—downstream burden reduction" cascade synergistic strategy through iterative search, demonstrating the practical benefits of incorporating the water quality cascade transmission relationship into the optimization objective calculation.

[0078] Step 3, based on the target configuration scheme output in Step 2, transforms the optimization results into actual constructed wetland engineering construction and mine water replenishment operation. During operation, water quality is continuously monitored. When the measured water quality deviates from the predicted value in Step 2, the scheme is dynamically updated until the water quality of all subsidence and water accumulation units meets the compliance requirements. The following uses the six subsidence and water accumulation unit examples from Steps 1 and 2 to explain the specific implementation process of Step 3.

[0079] First, constructed wetlands are built in each subsidence and water accumulation unit according to the target configuration scheme. The construction involves three aspects: substrate laying, plant planting, and inlet layout. All three are strictly carried out in accordance with the substrate type, planted plants, and inlet location corresponding to each subsidence and water accumulation unit in the target configuration scheme output in step 2.

[0080] The substrate laying method is as follows: In the shallow water area of ​​each subsidence and water accumulation unit or in the wetland treatment area formed by local cofferdams, an artificial wetland substrate layer is laid according to the substrate type determined by the target configuration scheme. The thickness of the substrate layer varies depending on the substrate type: the recommended thickness for zeolite substrate layer is 0.4m to 0.6m, with zeolite particle size selected in the range of 3mm to 8mm to balance cation exchange capacity and hydraulic permeability; the recommended thickness for gravel substrate layer is 0.3m to 0.5m, with gravel particle size selected in the range of 10mm to 30mm to ensure sufficient porosity and hydraulic conductivity; the recommended thickness for biochar substrate layer is 0.3m to 0.5m, with biochar particle size selected in the range of 1mm to 5mm. The biochar must be washed with water before laying to remove surface ash and soluble tar components to avoid releasing additional pollutants into the water body during the initial operation phase. In this embodiment, unit A has a 0.5m thick zeolite matrix layer, unit B has a 0.4m thick biochar matrix layer, unit C has a 0.4m thick biochar matrix layer, unit D has a 0.4m thick gravel matrix layer, unit E has a 0.5m thick zeolite matrix layer, and unit F has a 0.4m thick biochar matrix layer. A 0.1m to 0.15m thick graded crushed stone cushion layer is laid below the matrix layers as a support and flow equalization layer to prevent the upper matrix particles from sinking and embedding into the bottom mud, thus losing effective porosity. In an optional embodiment, the thickness of the matrix layer can be appropriately increased or decreased according to the actual water depth and foundation bearing capacity of each subsidence and water accumulation unit. For shallower subsidence and water accumulation units, the matrix layer can be thinned to reduce the amount of work, while for deeper subsidence and water accumulation units, a coarse gravel protective layer can be added above the matrix layer to resist water erosion.

[0081] Planting was carried out according to the plant types determined by the target configuration scheme. Reeds were planted by rhizome cuttings at a density of 9 to 16 plants per square meter, with row and plant spacing of 0.25 to 0.35 meters. The planting depth was ideally such that the top of the rhizome was 0.05 meters below the substrate surface. Planting was chosen during the spring months of March and April when water temperatures rose to facilitate rapid root development. Cattails were planted by transplanting tillers at a density of 6 to 12 plants per square meter, with row spacing of 0.3 to 0.4 meters. Cattails have strong root penetration and are highly adaptable to eutrophic water bodies, rapidly establishing colonies in subsidence areas with high total phosphorus concentrations. Sweet flag was planted by transplanting divisions at a density of 8 to 14 plants per square meter. The volatile terpenes in sweet flag have an inhibitory effect on some algae in the water, helping to maintain wetland water transparency. In this embodiment, unit A is planted with reeds at a density of 12 plants per square meter; unit B is planted with cattails at a density of 9 plants per square meter; unit C is planted with reeds at a density of 12 plants per square meter; unit D is planted with sweet flag at a density of 10 plants per square meter; unit E is planted with cattails at a density of 9 plants per square meter; and unit F is planted with cattails at a density of 9 plants per square meter. After planting, the water level in the wetland treatment area must be maintained stable at 0.05m to 0.15m above the substrate surface for 30 days to allow the plant roots to fully penetrate the substrate pores under shallow water conditions and establish a stable rhizosphere microbial film. In optional embodiments, the planting density can be flexibly adjusted according to the local seedling supply and the target colony establishment period. In scenarios where rapid results are urgently needed, the initial planting density can be increased; in cases of seedling shortage, the initial density can be decreased and the natural propagation period of the plants can be extended.

[0082] The inlet placement is implemented in each subsidence and water accumulation unit according to the inlet location determined by the target configuration scheme. In the long-side placement method, a horizontal water distribution pipe is installed at the end of one long side of the subsidence and water accumulation unit. The length of the water distribution pipe along the long side is no less than one-third of the total length of that long side of the subsidence and water accumulation unit. The pipe diameter is selected from DN150 to DN200. A 15mm diameter outlet hole is opened every 0.3m in the pipe wall. Mine water seeps evenly from the outlet holes of the water distribution pipe and forms a plug-flow hydraulic path from the inlet end to the opposite end within the unit. The hydraulic residence time is relatively long, suitable for long and narrow units to fully utilize the contact removal effect of the substrate and plants. In the long-side middle placement method, a water distribution pipe is installed at the center of the long side of the subsidence and water accumulation unit. Mine water diffuses to both sides simultaneously. This placement method divides the unit into two symmetrical treatment areas, making the hydraulic distribution more uniform in larger subsidence and water accumulation units. In the four-corner dispersed layout, one inlet pipe is set at each of the four corners of the subsidence water accumulation unit. The water output of each inlet is one-quarter of the total water intake of the subsidence water accumulation unit. Mine water is injected simultaneously from four directions and then converges in the central area of ​​the unit, minimizing hydraulic dead zones. This method is suitable for subsidence water accumulation units with large areas or near-square or irregular shapes. In this embodiment, unit A is arranged at one end of its long side, unit B at the middle of its long side, unit C at the four corners, unit D at one end of its long side, unit E at the middle of its long side, and unit F at the four corners. An overflow weir or perforated collection pipe is set on the outlet side of each subsidence water accumulation unit as the outlet. The outlet and inlet are kept as far apart as possible to extend the hydraulic residence path and reduce the short-circuit risk of mine water flowing out directly without sufficient treatment.

[0083] After the constructed wetland is completed, mine water is continuously pumped into each subsidence and water accumulation unit according to the mine water allocation ratio determined by the target configuration scheme. The main pipe of the mine water transmission and distribution network is led out from the mine water treatment station, and after sedimentation pretreatment, it is sent to the branch pipes of each subsidence and water accumulation unit through the main pipe. Electric valves and electromagnetic flow meters are installed on each branch pipe, and the opening degree of the electric valves is set according to the mine water allocation ratio determined by the target configuration scheme. In this embodiment, the total outflow of mine water is 12,000 cubic meters per day. The mine water allocation ratio of unit A is 20%, corresponding to an allocation flow of 2,400 cubic meters per day; unit B is 15%, corresponding to 1,800 cubic meters per day; unit C is 25%, corresponding to 3,000 cubic meters per day; unit D is 10%, corresponding to 1,200 cubic meters per day; unit E is 15%, corresponding to 1,800 cubic meters per day; and unit F is 15%, corresponding to 1,800 cubic meters per day. The total flow rate distributed across the six subsidence water accumulation units is 12,000 cubic meters per day, which is exactly equal to the total outflow of mine water. Electromagnetic flowmeters on each branch pipe monitor the actual flow rate in real time. When the actual flow rate deviates from the target distribution flow rate by more than ±5%, the field controller automatically adjusts the opening of the corresponding electric valve to compensate, ensuring that the actual mine water flow rate received by each subsidence water accumulation unit remains consistent with the mine water distribution ratio of the target configuration scheme.

[0084] After being pumped into each subsidence water unit, the mine water is purified by constructed wetlands before being replenished into the subsidence water unit. The purification process occurs simultaneously through three pathways: In the matrix adsorption pathway, as the mine water seeps through the pores of the matrix layer, dissolved ammonium ions in the water are captured by the cation exchange sites of zeolite, and phosphate ions are complexed and fixed by oxygen-containing functional groups on the surface of biochar, or intercepted by the biofilm on the gravel surface; In the plant absorption pathway, ammonia nitrogen is converted into nitrate nitrogen by rhizosphere nitrifying bacteria and then actively absorbed and assimilated into plant tissues by plant roots, while total phosphorus is transported across the membrane in the form of inorganic phosphate through plant roots to participate in plant metabolism; In the microbial degradation pathway, the microbial community in the biofilm attached to the surface of matrix particles and plant roots converts ammonia nitrogen into nitrogen gas through nitrification-denitrification processes and escapes from the water body, while polyphosphate-accumulating bacteria complete the excessive uptake and release cycle of phosphorus under alternating aerobic and anaerobic conditions. The synergistic effect of the three pathways allows the mine water treated by the constructed wetland to replenish the open water body of the subsidence water unit with lower ammonia nitrogen and total phosphorus concentrations after flowing out of the wetland treatment area, gradually improving the overall water quality of the subsidence water unit.

[0085] Once operational, the water quality of each subsidence water accumulation unit needs to be periodically monitored to determine whether the target configuration scheme needs to be dynamically updated. Specifically, the dissolved oxygen, ammonia nitrogen, and total phosphorus concentrations of each subsidence water accumulation unit are measured every 14 days. Sampling points are set 1m downstream of the outlet of each unit, at a depth of 0.3m to 0.5m below the water surface, to represent the actual water quality after wetland treatment before it flows into open water bodies. The 14-day monitoring cycle addresses two needs: firstly, the treatment effect of the constructed wetland on mine water requires a certain hydraulic retention time to be fully realized; too short an interval will result in the collection of transitional water samples before the treatment process is complete, leading to evaluation conclusions that deviate from the actual treatment capacity; secondly, too long an interval will fail to capture water quality deterioration trends in a timely manner, delaying adjustments. In an optional implementation, the monitoring period can be adjusted appropriately according to the season and climate conditions. In summer, when microbial activity is high and water quality changes rapidly, it can be shortened to 10 days, while in winter, when water temperature is low and biochemical reaction rate decreases, it can be extended to 21 days.

[0086] After each monitoring session, the measured ammonia nitrogen concentration of each subsidence area is compared with the predicted effluent ammonia nitrogen concentration of the corresponding subsidence area in step 2. Simultaneously, the measured total phosphorus concentration is compared with the corresponding predicted effluent total phosphorus concentration. The predicted effluent ammonia nitrogen concentration and predicted effluent total phosphorus concentration were determined during the target value calculation in step 2: for each subsidence area, the predicted effluent ammonia nitrogen concentration equals the influent ammonia nitrogen concentration minus the product of the influent ammonia nitrogen concentration and the comprehensive ammonia nitrogen removal coefficient; the predicted effluent total phosphorus concentration equals the influent total phosphorus concentration minus the product of the influent total phosphorus concentration and the comprehensive total phosphorus removal coefficient. These two predicted values ​​represent the expected effluent water quality of each subsidence area under ideal conditions according to the target configuration scheme.

[0087] When the ammonia nitrogen concentration of any subsidence water unit is higher than the predicted effluent ammonia nitrogen concentration of the corresponding subsidence water unit in step 2 for two consecutive measurements, or when the total phosphorus concentration is higher than the predicted effluent total phosphorus concentration of the corresponding subsidence water unit in step 2 for two consecutive measurements, it is determined that the actual treatment effect of the subsidence water unit has deviated from the expected target configuration scheme, and a scheme update needs to be triggered. The reason for requiring two consecutive measurements rather than just one is to eliminate the interference of random fluctuations in a single sampling. Short-term events such as rainstorm erosion, seasonal temperature changes, and algal blooms may cause abnormally high single monitoring data, but such disturbances usually subside on their own in the next monitoring cycle. If a scheme update is triggered based on only one exceedance, it will lead to frequent and unnecessary repeated optimization. Two consecutive exceedances mean that the deviation is persistent, indicating that the substrate-plant-flow configuration in the current target configuration scheme can no longer adapt to the actual water quality changes of the subsidence water unit, and it is indeed necessary to re-optimize.

[0088] The specific operations after triggering the update are as follows: update the field information in step 1 with the most recently measured dissolved oxygen, ammonia nitrogen, and total phosphorus concentrations, replacing the corresponding water quality parameter values ​​originally collected in step 1 for the subsidence water accumulation unit. Then, return to step 2 to re-execute the hybrid frog-leap algorithm coupled with the Pareto multi-objective optimization process. The reason for using the most recently measured data instead of the average of two consecutive measurements or other statistics is that the most recently measured data best represents the current true water quality status of the subsidence water accumulation unit. The earlier data is 14 days away from the current time, during which time the water quality may have changed further. Hydrological parameters (water accumulation area and average water accumulation depth) and the relationship between the surface elevation difference and surface runoff direction between adjacent subsidence water accumulation units usually do not change significantly in the short term. Therefore, the original recorded values ​​in step 1 can be used when updating the scheme. However, if engineers observe significant changes in the water area or average water depth of a subsidence water accumulation unit due to continuous rainfall or changes in groundwater level during operation, the updated hydrological parameters can also be written into the field information in step 1.

[0089] After returning to step 2, using the updated site information as input, the entire process of defining decision variables, calculating comprehensive removal coefficients, calculating optimization target values, Pareto sorting, and frog swarm iterative search is re-executed to generate an updated target configuration scheme. Compared to the initially generated target configuration scheme, the updated target configuration scheme may differ in the matrix type, planted vegetation, inlet location, or mine water distribution ratio of several subsidence and water accumulation units. In actual engineering, adjustments to the matrix type and planted vegetation require a certain construction period. Therefore, for subsidence and water accumulation units where the matrix type and planted vegetation have changed, partial renovations are necessary—removing the original matrix layer, laying new matrix, and replanting new vegetation. Adjusting the mine water distribution ratio is relatively convenient; the switch can be completed on the same day simply by modifying the opening setting value on the electric valves of each branch pipe in the distribution network. If the inlet location changes, a water distribution pipe needs to be added at the new location, and the old inlet needs to be sealed. In an optional implementation, to reduce construction costs and ecological disturbances caused by frequent substrate overhauls and plant replacements, additional constraints can be imposed on the optimization process in step 2 when updating the scheme: for subsidence water accumulation units that have not triggered updates, their substrate type and planted plants are locked and only the mine water distribution ratio and inlet location are allowed to participate in the optimization, so that the difference between the updated target configuration scheme and the previous scheme is concentrated on the variables with the lowest adjustment cost.

[0090] The following is an explanation of a hypothetical operating scenario in this embodiment. After the initial target configuration scheme was put into operation, two consecutive measured results on day 28 (i.e., the second monitoring) and day 42 (i.e., the third monitoring) showed that the ammonia nitrogen concentration in unit F was 4.8 mg / L and 5.1 mg / L, respectively, both higher than the predicted effluent ammonia nitrogen concentration of 3.2 mg / L calculated for unit F in step 2. The reason for this is that the average water depth of unit F is 2.2 m, making it the deepest of all subsidence water-accumulating units, and the dissolved oxygen concentration is only 1.8 mg / L. Although step 2 corrected for water depth and dissolved oxygen in unit F, the anaerobic conditions in the deeper water layers may have worsened further than at the time of data collection, resulting in an actual ammonia nitrogen removal effect lower than the corrected predicted value. After triggering the update, the measured dissolved oxygen concentration (1.2 mg / L), ammonia nitrogen concentration (5.1 mg / L), and total phosphorus concentration (1.1 mg / L) of Unit F on day 42 are updated to the field information in step 1. The process then returns to step 2 to re-execute the hybrid frog-leap algorithm coupled with Pareto multi-objective optimization. During the re-optimization process, since the ammonia nitrogen concentration in Unit F decreased from the initial 10.5 mg / L to 5.1 mg / L (indicating that the previous operation had a certain purification effect, but did not reach the predicted value), and the dissolved oxygen concentration decreased from 1.8 mg / L to 1.2 mg / L (anaerobic conditions worsened), the algorithm may make the following adjustments to Unit F in the updated target configuration: reduce the mine water allocation ratio from 15% to 10% to reduce the hydraulic load, and transfer the released 5% allocation ratio to Unit D, which still has a processing capacity surplus; or keep the mine water allocation ratio unchanged, but change the substrate type from biochar to zeolite to enhance the ion exchange adsorption capacity for ammonia nitrogen to compensate for the insufficient microbial nitrification capacity. The specific adjustment result is automatically determined by the algorithm based on the updated global optimization target value.

[0091] The dynamic updating process of the target configuration scheme can be repeated. After each update, the scheme is put into operation, and the water quality of each subsidence and water accumulation unit is monitored every 14 days. If any subsidence and water accumulation unit exceeds the standard twice consecutively, an update is triggered again. This closed loop of "operation-monitoring-evaluation-update-reoperation" continues until the ammonia nitrogen concentration and total phosphorus concentration of all subsidence and water accumulation units meet the Class IV water quality requirements of the Surface Water Environmental Quality Standard for six consecutive months. The Class IV water quality requirements of the Surface Water Environmental Quality Standard are: ammonia nitrogen concentration not exceeding 1.5 mg / L and total phosphorus concentration not exceeding 0.3 mg / L. The 6-month continuous compliance criterion means that within at least approximately 13 monitoring cycles (calculated at 14 days per cycle, 6 months is approximately 180 days, 180 divided by 14 is approximately 13 cycles), the measured ammonia nitrogen concentration in all subsidence water units must not exceed 1.5 mg / L in any single measurement, and the measured total phosphorus concentration must not exceed 0.3 mg / L in any single measurement. Setting a 6-month continuous compliance period takes into account the seasonal fluctuations in water quality. A 6-month span is sufficient to cover at least two complete seasonal transitions (e.g., from spring to autumn or from summer to winter), ensuring that the compliance conclusion is not only valid in a specific season but can be maintained under different temperature, rainfall, and evaporation conditions. In optional implementations, the continuous compliance period can be appropriately adjusted according to actual engineering needs and the seasonal fluctuations in water quality. For example, it can be shortened to 4 months in areas with smaller water quality fluctuations, and extended to 8 or 12 months in high-latitude areas with drastic seasonal changes to ensure coverage of the complete annual climate cycle.

[0092] Once the ammonia nitrogen and total phosphorus concentrations in all subsidence water accumulation units meet the aforementioned standards for six consecutive months, the collaborative restoration work of mine water ecological replenishment wetlands in the subsidence water accumulation area is completed. At this point, the water bodies in each subsidence water accumulation unit have transformed from their original state of excessive ammonia nitrogen and total phosphorus concentrations to ecological water bodies that meet the Class IV surface water quality standards. Mine water is no longer discharged as a pollution source, but is transformed into an ecological replenishment source for the subsidence water accumulation area after purification through artificial wetlands. The water bodies in the subsidence area have been improved in both quantity and quality. The plant communities in the artificial wetlands also provide vegetation cover and landscape restoration functions for the subsidence area, realizing the synergy between the resource utilization of mine water and the ecological restoration of the subsidence area.

[0093] In an optional implementation, the operation and maintenance phase after achieving the target can switch to a low-frequency monitoring mode, extending the monitoring cycle from every 14 days to every 30 days, with intensified monitoring only after extreme weather events (such as continuous heavy rain or the end of a severe winter ice-covered period). The continuous pumping flow rate of mine water can be adjusted synchronously according to the seasonal changes in the actual mine inflow: when the mine inflow increases during the wet season, the pumping flow rate of each subsidence water accumulation unit is increased proportionally; when the inflow decreases during the dry season, it is decreased proportionally, and the relative distribution ratio of mine water among each subsidence water accumulation unit remains unchanged. After the above-ground parts of the plants in the constructed wetland wither in autumn and winter, they need to be harvested and removed to prevent secondary pollution from the decomposition of litter. When harvesting, the root stubble height should be left at least 0.1m to protect overwintering buds and ensure the natural recovery of the plant community in the following spring. After 3 to 5 years of continuous operation, the substrate layer may experience adsorption capacity saturation or pore blockage, and the need for substrate replacement or regeneration should be assessed based on the operation monitoring data.

[0094] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result according to substantially the same method falls within the scope of the present invention. Therefore, the scope of the present invention is defined only by the appended claims.

Claims

1. A method for the coordinated restoration of mine water ecological replenishment wetlands in subsidence and waterlogged areas, characterized in that, Includes the following steps: Step 1: Divide the coal mining subsidence and water accumulation area into several subsidence and water accumulation units according to the surface elevation boundary, collect the water quality and hydrological parameters of each subsidence and water accumulation unit, record the relationship between the surface elevation difference and surface runoff direction between adjacent subsidence and water accumulation units, and record the total flow rate of mine water outflow. Step 2: For each subsidence water accumulation unit, determine the decision variables including substrate type, planted vegetation, inlet location, and mine water allocation ratio. Concatenate the decision variables of all subsidence water accumulation units into a decision vector and define it as one frog. Randomly generate several frogs to form an initial population. The sum of the mine water allocation ratios of all subsidence water accumulation units in each frog is equal to 100%. Based on the decision vector of each frog and the field information collected in Step 1, calculate the comprehensive removal coefficient for each unit and calculate multiple optimization target values ​​corresponding to each frog accordingly. Perform Pareto sorting on all frogs in the population and distribute them to several frog groups at equal intervals. Within each frog group, update the decision vector of the worst frog in the local group based on the best frog in the local group and determine whether to replace it based on the Pareto dominance relationship. After completing the local search of all frog groups, mix all frogs, re-execute Pareto sorting and grouping, and enter the next iteration. After the iteration terminates, select the target configuration scheme from the Pareto front set. Step 3: Construct artificial wetlands in each subsidence and water accumulation unit according to the target configuration scheme, based on the determined substrate type, plant species, and inlet location. Continuously pump mine water into each subsidence and water accumulation unit according to the determined mine water distribution ratio. After purification by the artificial wetlands, the mine water is replenished into the water body of the subsidence and water accumulation unit, thus completing the ecological replenishment wetland collaborative restoration of mine water in the subsidence and water accumulation area. In step 2, the substrate type is selected from zeolite, gravel, and biochar; the planting plant is selected from reeds, cattails, and calamus; the inlet location is selected from one end of the long side, the middle of the long side, and the four corners; the mine water distribution ratio is the percentage of the mine water flow allocated to the corresponding subsidence water accumulation unit relative to the total mine water outflow, with a step size of 5% and a range of 5% to 60%. In step 2, the method for calculating the comprehensive removal coefficient for each unit is as follows: the matrix ammonia nitrogen removal coefficient is determined by matrix type, with zeolite taking 0.60, biochar taking 0.50, and gravel taking 0.30; the matrix total phosphorus removal coefficient is determined by matrix type, with biochar taking 0.60, zeolite taking 0.40, and gravel taking 0.20; the plant ammonia nitrogen removal coefficient is determined by plant species, with reed taking 0.30, cattail taking 0.20, and calamus taking 0.15; the plant total phosphorus removal coefficient is determined by plant species, with cattail taking 0.30, reed taking 0.20, and calamus taking 0.15; the matrix ammonia nitrogen removal coefficient and the plant ammonia nitrogen removal coefficient are added together to obtain the comprehensive ammonia nitrogen removal coefficient, and the matrix total phosphorus removal coefficient and the plant total phosphorus removal coefficient are added together to obtain the comprehensive total phosphorus removal coefficient. The upper limit for each comprehensive removal coefficient is 0.

90. For adjacent subsidence aquifers with a surface elevation difference greater than 0.5m and connected surface runoff directions, the effluent ammonia nitrogen concentration of the upstream subsidence aquifer is first calculated by subtracting the product of the influent ammonia nitrogen concentration and the comprehensive ammonia nitrogen removal coefficient of the upstream subsidence aquifer. The influent of the downstream subsidence aquifer includes both the effluent from the upstream subsidence aquifer and the mine water directly received from the mine water network by the downstream subsidence aquifer. The effluent ammonia nitrogen concentration is calculated by adding the product of the effluent ammonia nitrogen concentration and the flow rate of the upstream subsidence aquifer. The influent ammonia nitrogen concentration of the downstream subsidence water-collecting unit is obtained by dividing the sum of the product of the ammonia nitrogen concentration and the distribution flow of the downstream subsidence water-collecting unit by the sum of the outflow flow of the upstream subsidence water-collecting unit and the distribution flow of the downstream subsidence water-collecting unit. The influent total phosphorus concentration of the downstream subsidence water-collecting unit is calculated in the same way. The outflow flow of any subsidence water-collecting unit is equal to the sum of its own distribution flow and the outflow flow from the upstream subsidence water-collecting unit. The outflow flow of the upstreamst subsidence water-collecting unit is equal to its own distribution flow. The series relationship is passed down step by step from the upstreamst subsidence water-collecting unit according to the surface runoff direction.

2. The method according to claim 1, characterized in that, In step 1, the water quality parameters collected for each subsidence water accumulation unit include dissolved oxygen concentration, ammonia nitrogen concentration and total phosphorus concentration, and the hydrological parameters include water accumulation area and average water accumulation depth.

3. The method according to claim 1, characterized in that, On-site corrections were performed based on the comprehensive removal coefficients for ammonia nitrogen and total phosphorus: For subsidence water units with dissolved oxygen concentrations below 2 mg / L, the comprehensive removal coefficient for ammonia nitrogen was multiplied by 0.60 to obtain the corrected comprehensive removal coefficient for ammonia nitrogen; for subsidence water units with an average water depth exceeding 1.5 m, the plant ammonia nitrogen removal coefficient and the plant total phosphorus removal coefficient were each multiplied by 0.80 and then recalculated in the comprehensive removal coefficient calculation; the two corrections were performed cumulatively.

4. The method according to claim 1, characterized in that, Step 2 includes three optimization objectives: The first objective is to maximize the total reduction of ammonia nitrogen load in the system, which is obtained by multiplying the distribution flow rate of each subsidence water unit by the influent ammonia nitrogen concentration and then by the ammonia nitrogen comprehensive removal coefficient, and then summing the results over all subsidence water units. The second objective is to maximize the total phosphorus load reduction of the system, which is obtained by multiplying the allocated flow rate of each subsidence water unit by the influent total phosphorus concentration and then by the total phosphorus comprehensive removal coefficient, and then summing the results for all subsidence water units. The third objective is to maximize the number of predicted compliant units, which is obtained by subtracting the product of the influent ammonia nitrogen concentration and the ammonia nitrogen comprehensive removal coefficient from the influent ammonia nitrogen concentration of each subsidence water unit to obtain the predicted effluent ammonia nitrogen concentration, and by subtracting the product of the influent total phosphorus concentration and the total phosphorus comprehensive removal coefficient from the influent total phosphorus concentration to obtain the predicted effluent total phosphorus concentration. Subsidence water units with predicted effluent ammonia nitrogen concentration not exceeding 1.5 mg / L and predicted effluent total phosphorus concentration not exceeding 0.3 mg / L are counted as compliant units.

5. The method according to claim 4, characterized in that, The initial population contains 100 frogs, which are equally distributed into 10 frog groups. Within each frog group, the frog with the largest sum of the first, second, and third objectives is defined as the locally optimal frog, and the frog with the smallest sum of the first, second, and third objectives is defined as the locally worst frog. For each decision variable of the locally worst frog, a random number between 0 and 1 is generated. The random number is multiplied by the difference between the locally optimal and locally worst frogs on the corresponding decision variables. The product is then added to the corresponding decision variable of the locally worst frog to generate a new candidate frog. If, after the update, the sum of the mine water allocation ratios of all subsidence water accumulation units is not equal to 100%, the deviation is distributed proportionally according to the existing mine water allocation ratios of each subsidence water accumulation unit and rounded up by a step size of 5%.

6. The method according to claim 5, characterized in that, If a candidate frog Pareto dominates the local worst frog, then the candidate frog replaces the local worst frog. If a candidate frog does not Pareto dominate the local worst frog, then one frog is randomly selected from the Pareto front set to replace the local best frog, and the update operation is repeated. If the frog generated after replacement still does not Pareto dominate the local worst frog, then one frog that satisfies the total constraint that the sum of the mine water distribution ratios of all subsidence water accumulation units equals 100% is randomly regenerated to replace the local worst frog. The iteration ends when the number of frogs in the Pareto front set and the three optimization objective values ​​of each frog no longer change after 20 consecutive iterations. The decision vector corresponding to the frog with the largest third objective is selected from the Pareto front set as the objective configuration scheme. If multiple frogs have the same third objective, the decision vector corresponding to the frog with the largest sum of the first and second objectives is selected as the objective configuration scheme.

7. The method according to claim 4, characterized in that, In step 3, the dissolved oxygen concentration, ammonia nitrogen concentration, and total phosphorus concentration of each subsidence water unit are measured once every 14 days. If the ammonia nitrogen concentration of any subsidence water unit is higher than the predicted effluent ammonia nitrogen concentration of the corresponding subsidence water unit in step 2 for two consecutive measurements, or if the total phosphorus concentration is higher than the predicted effluent total phosphorus concentration of the corresponding subsidence water unit in step 2 for two consecutive measurements, the most recently measured dissolved oxygen concentration, ammonia nitrogen concentration, and total phosphorus concentration are updated to the field information in step 1, and the process is returned to step 2 to re-execute the hybrid frog-jumping algorithm and Pareto multi-objective optimization coupling process to generate an updated target configuration scheme until the ammonia nitrogen concentration and total phosphorus concentration of all subsidence water units meet the Class IV water quality requirements of the surface water environmental quality standard for 6 consecutive months.

Images