Optimization method for multistage PRB in-situ remediation of acid mine water of coal mine

Abstract

The invention discloses a coal mine acid mine water multistage PRB in-situ remediation optimization method, and relates to the technical field of numerical simulation. Comprising the following steps: constructing a two-dimensional geometric model for treating the acid mine water of the coal mine through multistage PRB; assigning the size and hydrogeological parameters of each stage of PRB filler to obtain a numerical simulation model; performing batch processing on the numerical simulation model to generate a plurality of groups of multi-stage PRB initial design parameters, and inputting the multi-stage PRB initial design parameters into the optimized numerical model for operation to obtain a plurality of objective function values; predicting each objective function value through a back-propagation neural network model, and taking the back-propagation neural network model with the best prediction effect as a substitution model of a numerical simulation model; and coupling the substitution model with a fast non-dominated sorting genetic algorithm to obtain a Pareto optimal solution set. The multi-stage PRB optimization method solves the problem that the prior art lacks clear guidance principles and evaluation standards, and improves the accuracy of the multi-stage PRB optimization method in practical engineering application.

Description

Technical Field

[0001] The invention relates to the technical field of numerical simulation, and in particular to an optimization method for in-situ remediation of acid mine water in coal mines using multi-stage PRBs. Background Art

[0002] my country has rich coal resources with high sulfur content. After the ore seams are exposed due to mining, high-sulfur coal will react in the coal seams to produce acid mine water. Acid mine water is a kind of polluted groundwater, which deeply harms the ecological environment and human health and safety. Multi-stage PRB (Permeable Reactive Barriers, in-situ passive remediation technology) as a method of in-situ remediation of mine water is currently a hot technology in the field of mine water pollution remediation. Setting multi-stage PRB in coal mine caves can effectively treat Fe and SO in acid mine water. 4 2- and other characteristic pollutants to improve the water quality of mine water.

[0003] At present, there is still a lack of multi-objective optimization methods for multi-stage PRB to treat acid mine water in high-sulfur coal mines. Most optimization methods are based on statistical methods, such as response surface method. There are relatively few theoretical studies on multi-objective optimization of multi-stage PRB, and a systematic theoretical framework and methodology have not yet been formed, which leads to a lack of clear guiding principles and evaluation criteria in practical applications; and the model construction is difficult, and there is a lack of models that meet actual needs and are easy to solve. Therefore, according to current research, it is difficult to meet the accurate and efficient multi-stage PRB optimization methods required in practical engineering applications. Summary of the invention

[0004] Based on this, it is necessary to provide an optimization method for multi-stage PRB in-situ remediation of acid mine water in coal mines in response to the above technical problems.

[0005] The embodiment of the present invention provides an optimization method for multi-stage PRB in-situ remediation of acid mine water in coal mines, comprising: Obtain the mine size parameters in the study area, determine the multi-stage PRB numerical simulation boundary and structure according to the mine size parameters, and construct a two-dimensional geometric model of multi-stage PRB treatment of acid mine water in coal mines; assign values ​​to the size and hydrogeological parameters of each level of PRB fillers in the two-dimensional geometric model, and construct a multi-stage PRB treatment of acid mine water numerical simulation model that simulates the migration and diffusion of characteristic pollutants of acid mine water in coal mines in each level of PRB fillers; and perform batch processing on the numerical simulation model to generate multiple sets of multi-stage PRB initial design parameters; An optimization numerical model was constructed with the service life and pollutant treatment capacity of the multi-stage PRB as the objective function and the hydrogeological parameters and concentration load of characteristic pollutants of the multi-stage PRB filler as the constraints. Inputting multiple sets of multi-level PRB initial design parameters into the optimization numerical model for calculation, and obtaining multiple objective function values ​​corresponding to the multiple sets of multi-level PRB initial design parameters; predicting each objective function value through a back propagation neural network model BPNN, and obtaining a prediction result corresponding to the objective function value; The prediction effect is evaluated by the correlation coefficient between each objective function value and the prediction result corresponding to the objective function value, and the back propagation neural network model BPNN with the best prediction effect is used as a substitute model for the numerical simulation model; The substitution model of the numerical simulation model is coupled with the fast non-dominated sorting genetic algorithm NSGA-II to obtain the Pareto optimal solution set of multi-stage PRB treatment of acid mine water in coal mines. The optimal multi-stage PRB design scheme for in-situ remediation of acid mine water in coal mines in the study area is selected from the Pareto optimal solution set.

[0006] Optionally, a numerical simulation model for treating acid mine water with multi-stage PRB is constructed to simulate the migration and diffusion of characteristic pollutants of acid mine water in each stage of PRB fillers, specifically including: Using limestone, coconut shell biochar, and anion exchange resin D201 as multi-stage PRB fillers, the hydrogeological parameters of each stage of PRB fillers are obtained; the hydrogeological parameters include filler porosity, permeability, dispersivity, and chemical reaction parameters; In the numerical simulation software, the porosity, permeability, dispersivity and chemical reaction parameters of each level of PRB filler are assigned, and the groundwater seepage model and reaction migration model of acid mine water in each filling material of multi-level PRB are constructed. The migration and diffusion of characteristic pollutants of acid mine water in coal mines in each level of filler are simulated to obtain a numerical simulation model.

[0007] Optionally, the numerical simulation model includes a groundwater seepage model and a reaction migration model; the groundwater seepage model has the formula: ; in, K is the permeability coefficient of the aquifer, H is the water level of the aquifer, S s is the water storage rate of the porous medium, t For time, x and y All are independent variables; Reaction transport model, its formula is: ; in, c is the solute concentration, D L is the hydrodynamic diffusion tensor, uis the actual groundwater flow velocity, c p is the concentration of pollutant adsorbed per unit solid dry weight, ρ b is the dry density of the solid medium, ρ b c p is the mass of solute adsorbed on the porous medium; k b c p is the chemical reaction mass of the solute in the porous medium.

[0008] Optionally, the service life and pollutant treatment capacity of the multi-stage PRB are used as the objective function, and the hydrogeological parameters of the multi-stage PRB filler and the concentration load of characteristic pollutants are used as constraints, specifically including: To maximize the effective service life of multi-stage PRB f 1 , maximize the effective water treatment volume of multi-stage PRB f 2 , maximize the amount of Fe treated by multi-stage PRB f 3 , maximize multi-stage PRB processing SO 4 2- Amount f 4 As the objective function, the length of limestone L PRB1 , biochar length L PRB2 , anion exchange resin length L PRB3 , initial water head H in , limestone porosity n PRB1 , anion exchange resin porosity n PRB2 , biochar porosity n PRB3 , Limestone Permeability K PRB1 , anion exchange resin permeability coefficient K PRB2 , Biochar permeability coefficient K PRB3 , Dispersion α , Influent Fe concentration c Fe 、Influent SO 4 2- concentration c SO4 2- As the constraint condition, the calculation formula is as follows: Max f 1 = min( D Fe , D SO4 2- ); Max f 2 = v × f 1 × A ; Max f 3 = f 2 × c Fe ; Max f 4 = f 2 × c SO4 2- ; st{∑ L PRBi = 6; L PRBi , n PRBi , K PRBi , H in , c Fe ∈0.01×N; α , c SO4 2- ∈0.1×N, i = 1,2,3}; in, D Fe and D SO4 2- For the i The time taken for a characteristic pollutant from the start of the simulation to the time when the concentration at the outlet exceeds the limit of the groundwater quality standard. v is the Darcy velocity at the outlet, A is the cross-sectional area perpendicular to the water flow direction, c Fe is the influent Fe concentration, c SO4 2- For influent SO 4 2- concentration, N is the reference value set.

[0009] Optionally, the prediction effect is evaluated by the correlation coefficient between the objective function value and the prediction result, specifically including: Construct a back propagation neural network model BPNN in Matlab, and divide the samples into training set and test set according to the set ratio; The training set is input into the back propagation neural network model BPNN for training, and the test set is input into the trained back propagation neural network model BPNN for prediction; Calculate the determination coefficient R between the prediction results and the numerical simulation results 2 , root mean square error RMSE, mean absolute error MAE and mean bias error MBE to evaluate the prediction effect of the back propagation neural network model BPNN; The back propagation neural network model BPNN with the best prediction effect is saved and used as a substitute model for the numerical simulation model.

[0010] Optionally, the surrogate model of the numerical simulation model is coupled with a fast non-dominated sorting genetic algorithm NSGA-II, specifically including: The Latin hypercube sampling method LHS in Matlab software was used to generate the initial population; The initial population is input into the surrogate model of the numerical simulation model for prediction, and the corresponding objective function is output. The Pareto frontier of the objective function is calculated by the fast non-dominated sorting genetic algorithm NSGA-II, and the Pareto optimal solution set of multi-stage PRB for treating acid mine water in coal mines is output.

[0011] The above-mentioned optimization method for multi-stage PRB in-situ remediation of acid mine water in coal mines provided by the embodiment of the present invention has the following beneficial effects compared with the prior art: In the process of constructing the optimization numerical model of the present invention, the hydrogeological parameters of the most representative multi-stage PRB fillers and the concentration loads of characteristic pollutants in the multi-stage PRB in-situ remediation process are selected as constraints, which are directly related to the performance of the PRB fillers and the removal effect of pollutants in mine water, accurately reflecting the actual situation of mine water treatment, solving the problem that the prior art lacks clear guiding principles and evaluation standards, and improving the accuracy of the multi-stage PRB optimization method in actual engineering applications.

[0012] In addition, the present invention introduces a back propagation neural network model BPNN to predict the target function value. The back propagation neural network model BPNN learns and simulates complex functional relationships to achieve rapid prediction, avoiding the problems of large amount of calculation and long calculation time faced by traditional simulation-optimization methods, and has wider applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1A flowchart of an optimization method for multi-stage PRB in-situ remediation of acid mine water in coal mines provided in one embodiment; Figure 2 The objective function for predicting the multi-stage PRB in-situ remediation of acid mine water in a coal mine provided in an embodiment is f 1 Linear fitting result diagram of the value and numerical simulation value; Figure 3 The objective function for predicting the multi-stage PRB in-situ remediation of acid mine water in a coal mine provided in an embodiment is f 2 The linear fitting results of the values ​​and numerical simulation values; Figure 4 A schematic diagram of the results of an optimization method for multi-stage PRB in-situ remediation of acid mine water in coal mines provided in an embodiment. DETAILED DESCRIPTION

[0014] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0015] The present invention solves the problem of the lack of clear guiding principles and evaluation criteria in the prior art, and avoids the limitation of single-objective optimization. A set of optimal solutions are calculated, which meet the requirements under multiple constraints, providing more abundant options for the design of multi-stage PRB processing technology. By introducing the back propagation neural network model BPNN, the problems of large amount of calculation and long calculation time faced by the traditional simulation-optimization method are avoided, so that it has a wider applicability.

[0016] In one embodiment, a method for optimizing the in-situ remediation of acid mine water in coal mines using multi-stage PRB is provided, such as Figure 1 As shown, the method includes: The mine size parameters in the study area are obtained, and the multi-stage PRB numerical simulation boundary and structure are determined according to the mine size parameters to construct a two-dimensional geometric model of multi-stage PRB treatment of acid mine water in coal mines. The size and hydrogeological parameters of each level of PRB fillers in the two-dimensional geometric model are assigned to construct a numerical simulation model of multi-stage PRB treatment of acid mine water in coal mines that simulates the migration and diffusion of characteristic pollutants of acid mine water in coal mines in each level of PRB fillers. And the numerical simulation model is batch processed to generate multiple sets of initial design parameters of multi-stage PRB.

[0017] An optimization numerical model was constructed with the hydrogeological parameters of the multi-stage PRB filler and the concentration load of characteristic pollutants as constraints, and the service life and pollutant treatment capacity of the multi-stage PRB as objective functions.

[0018] Multiple sets of multi-level PRB initial design parameters are input into the optimization numerical model for calculation, and multiple objective function values ​​corresponding to the multiple sets of multi-level PRB initial design parameters are obtained. Each objective function value is predicted by the back propagation neural network model BPNN (Back Propagation Neural Network) to obtain the prediction results corresponding to the objective function value.

[0019] The prediction effect is evaluated by the correlation coefficient between each objective function value and the prediction result corresponding to the objective function value, and the back propagation neural network model BPNN with the best prediction effect is used as a substitute model for the numerical simulation model.

[0020] The substitution model of the numerical simulation model is coupled with the fast non-dominated sorting genetic algorithm NSGA-II to obtain the Pareto optimal solution set of multi-stage PRB treatment of acid mine water in coal mines. The optimal multi-stage PRB design scheme for in-situ remediation of acid mine water in coal mines in the study area is selected from the Pareto optimal solution set.

[0021] The specific implementation process includes: S1 obtains the mine size parameters in the study area, designs the multi-stage PRB numerical simulation boundary and structure, constructs a two-dimensional geometric model of multi-stage PRB for treating acid mine water in coal mines, assigns the size and hydrogeological parameters of each stage of PRB filler in Comsol (COMSOL Multiphysics, multi-physics field simulation software) software, and generates a numerical model of multi-stage PRB for treating acid mine water in coal mines.

[0022] The numerical simulation model of multi-stage PRB for treating high-sulfur coal mine acid mine water provided by the present invention includes a groundwater seepage model and a reaction migration model. The specific construction steps include: The mine size parameters in the study area were obtained, the multi-stage PRB numerical simulation boundary and structure were designed, and a two-dimensional geometric model of multi-stage PRB for treating acid mine water in coal mines was constructed. The size of each PRB filler was assigned in Comsol software to generate a multi-stage PRB geometric model.

[0023] Limestone, coconut shell biochar and anion exchange resin D201 were selected as multi-level PRB filling materials, and the relevant hydrogeological parameters of each level of PRB filling were obtained, including filling porosity, permeability coefficient, dispersivity and chemical reaction parameters. The filling porosity, permeability coefficient, dispersivity and chemical reaction parameters of each level of PRB filling were assigned in the numerical simulation software, and the groundwater seepage model and reaction migration model of acid mine water in coal mines in each filling material of multi-level PRB were constructed to simulate the migration and diffusion of characteristic pollutants of acid mine water in coal mines in each level of filling, and obtain the numerical simulation model.

[0024] Among them, the groundwater seepage model is: ; Where: K is the permeability coefficient of the aquifer, m / d; H is the water level elevation of the aquifer, m; S s is the water storage rate of the porous medium, 1 / m, i.e., the amount of water released when the aquifer drops one unit head; t is the time, d (d represents day); x and y All are independent variables.

[0025] Reaction Transport Model: ; Where: c is the solute concentration, mg / L; D L is the hydrodynamic diffusion tensor, m 2 / d; u is the actual groundwater flow velocity, m / d; c p is the concentration of pollutant adsorbed per unit solid dry weight, mg / kg; ρ b is the dry density of solid medium, kg / m 3 , ρ b c p is the mass of solute adsorbed on the porous medium; k b c p is the chemical reaction mass of the solute in the porous medium.

[0026] S2 saved the constructed multi-stage PRB numerical simulation model for treating acid mine water in coal mines, connected to Comsol using the comsol withmatlab interface, wrote code using Matlab (Matrix Laboratory), generated 500 sets of multi-stage PRB initial design parameters, input the saved Comsol model for calculation, and output the corresponding objective function value, thus realizing batch processing of the numerical simulation model.

[0027] S3 The service life and pollutant treatment capacity of the multi-stage PRB were selected as the objective function, and the hydrogeological parameters of the multi-stage PRB filler and the concentration load of characteristic pollutants were selected as constraints to construct an optimization numerical model.

[0028] The relevant parameters of the constraints and objective function include: To maximize the effective service life of multi-stage PRB ( f 1 , unit: d), maximize the effective water treatment volume of multi-stage PRB ( f 2 , unit: m 3 ), maximize the amount of Fe treated by multi-stage PRB ( f 3 , unit: kg), maximize multi-stage PRB processing SO 4 2- The amount ( f 4 , unit: kg) as the objective function, with limestone length ( L PRB1 , unit: m), biochar length ( L PRB2 , unit: m), anion exchange resin length ( L PRB3 , unit: m), initial water head ( H in , unit: m), limestone porosity ( n PRB1 , dimensionless), anion exchange resin porosity ( n PRB2 , dimensionless), biochar porosity ( n PRB3 , dimensionless), limestone permeability coefficient ( K PRB1 , unit: m / d), anion exchange resin permeability coefficient ( K PRB2 , unit: m / d), biochar permeability coefficient ( K PRB3 , unit: m / d), dispersivity ( α , unit m), influent Fe concentration ( c Fe , unit: mol / m 3 )、Influent SO 4 2- concentration( c SO4 2- , unit: mol / m 3 ) is a constraint condition, and the calculation formula is as follows: Max f 1 = min( D Fe , D SO4 2- ); Max f 2 =v × f 1 × A ; Max f 3 = f 2 × c Fe ; Max f 4 = f 2 × c SO4 2- ; st{∑ L PRBi = 6; L PRBi , n PRBi , K PRBi , H in , c Fe ∈0.01×N; α , c SO4 2- ∈0.1×N, i = 1,2,3}; Where: D Fe and D SO4 2- For the i The time taken for the characteristic pollutant to exceed the groundwater quality standard limit from the beginning of the simulation to the concentration at the outlet (Fe is 2 mg / L, SO 4 2- is 350 mg / L), unit: d; v is the Darcy velocity at the outlet, in m / d; A is the cross-sectional area perpendicular to the water flow direction, in m 2 ; c Fe and c SO4 2- Influent Fe and SO 4 2- Concentration, unit mol / m 3 , N is the reference value set.

[0029] like Figure 2 As shown, Figure 2 The objective function for prediction f1 The linear fitting result of (maximizing the effective service life of multi-level PRB) value and numerical simulation value. The horizontal axis in the figure is the numerical simulation value, and the vertical axis is the BPNN model prediction value. The color scale in the figure represents the point density, which can characterize the concentration of the data. It can be seen from the figure that f 1 The determination coefficient R between the numerical simulation value and the predicted value 2 The error indexes such as MAE, MBE and RMSE are low, indicating that the machine learning model is f 1 The prediction is more accurate.

[0030] like Figure 3 As shown, Figure 3 The objective function for prediction f 2 The linear fitting results of (maximizing the effective water treatment volume of multi-stage PRB) values ​​and numerical simulation values. The horizontal axis in the figure is the numerical simulation value, and the vertical axis is the BPNN model prediction value. The color scale in the figure represents the point density, which can characterize the concentration of the data. As can be seen from the figure, f 2 The determination coefficient R between the numerical simulation value and the predicted value 2 The error indexes such as MAE, MBE and RMSE are low, indicating that the machine learning model is f 2 The prediction is more accurate.

[0031] S4 uses the back propagation neural network model BPNN (Back Propagation Neural Network) to predict the output results of the optimized numerical model, calculate the predicted results for evaluation, perform multiple calculations, and save the neural network with the best prediction effect.

[0032] The multi-objective optimization results of the multi-stage PRB treatment of high-sulfur coal mine acid mine water provided by the present invention include the following steps: The calculated results in S3 are saved for subsequent prediction. The back propagation neural network model BPNN is constructed in Matlab, and 500 samples are divided into training set and test set according to the set ratio (8:2). The saved results are predicted, and the determination coefficient R between the predicted results and the numerical simulation results is calculated. 2 , root mean square error RMSE, mean absolute error MAE and mean bias error MBE to evaluate the prediction effect of the back propagation neural network model BPNN. By debugging the hyperparameters of the back propagation neural network model BPNN multiple times, the back propagation neural network model BPNN with the best prediction effect is saved and used as a substitute model for the numerical simulation model.

[0033] like Figure 4 As shown, Figure 4 The 500 Pareto optimal solution sets calculated by the trained BPNN model are represented in the form of parallel coordinate graphs. f 1 — f 4 ( f 1 To maximize the effective service life of multi-stage PRB, f 2 To maximize the effective water treatment capacity of multi-stage PRB, f 3 To maximize the amount of Fe processed by multi-stage PRB, f 4 To maximize the multi-stage PRB processing SO 4 2- ) represent the values ​​of the four objective functions, among which f 1 The value range is 1440~5235; f 2 The value range is 910~1390; f 3 The value range is 19~42; f 4 The value range of is 1428~3586. Each curve in the figure represents the objective function value corresponding to each Pareto optimal solution, and the color of the curve represents f 1 The color scale of , according to the color of the scale, the correlation relationship between the objective functions can be judged, for example f 1 The bigger, f 2 The smaller. Figure 4 It can be seen that the trained BPNN model predicts the value R of the objective function 2 They are 0.986 and 0.993, respectively, indicating that the model has good prediction effect and can be used as an alternative model for multi-level PRB numerical simulation model.

[0034] S5 uses the Latin hypercube sampling method LHS in Matlab software to generate the initial population, inputs it into the alternative model of the numerical simulation model (the saved back propagation neural network model BPNN) for prediction, outputs the corresponding objective function value, and calculates the Pareto frontier of the objective function value through the fast non-dominated sorting genetic algorithm NSGA-II, and outputs the Pareto optimal solution set (Pareto Optimal Set, also known as the Pareto optimal set) of the objective function.

[0035] Set the initial population to 500 and the number of iterations to 1000, perform a fast non-dominated sort on the population and calculate the corresponding crowding degree, and generate the offspring population through selection, crossover, and mutation operations. Repeat the above steps until the specified number of iterations is reached, and output the Pareto optimal solution set of the objective function. Save the calculated Pareto optimal solution set of the objective function, and use a parallel coordinate graph to represent the predicted objective function value.

[0036] The above-mentioned embodiments only express several implementation methods of the present invention, and the description is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. An optimization method for multi-stage PRB in-situ remediation of acid mine water in coal mines, characterized in that: include: According to the mine size parameters in the study area, a numerical simulation model of multi-stage PRB for treating acid mine water in coal mines was constructed; The numerical simulation model is batch processed to generate multiple sets of multi-level PRB initial design parameters; An optimization numerical model was constructed with the service life and pollutant treatment capacity of the multi-stage PRB as the objective function and the hydrogeological parameters and concentration load of characteristic pollutants of the multi-stage PRB filler as the constraints. Input multiple groups of multi-level PRB initial design parameters into the optimization numerical model for operation to obtain multiple objective function values; predict each objective function value through the back propagation neural network model to obtain the prediction result; evaluate the prediction effect through the correlation coefficient between each objective function value and the prediction result, and use the back propagation neural network model with the best prediction effect as a substitute model for the numerical simulation model; The substitution model of the numerical simulation model is coupled with the fast non-dominated sorting genetic algorithm to obtain the Pareto optimal solution set of multi-stage PRB treatment of acid mine water in coal mines. The optimal multi-stage PRB design scheme for multi-stage PRB in-situ remediation of acid mine water in coal mines in the study area is selected from the Pareto optimal solution set.

2. The optimization method for multi-stage PRB in-situ remediation of acid mine water in coal mines according to claim 1, characterized in that: According to the mine size parameters in the study area, a numerical simulation model of multi-stage PRB treatment of acid mine water in coal mines is constructed, which specifically includes: The numerical simulation boundary and structure of multi-stage PRB are determined according to the mine size parameters to construct a two-dimensional geometric model of multi-stage PRB for treating acid mine water in coal mines. The sizes and hydrogeological parameters of each level of PRB fillers in the two-dimensional geometric model are assigned, and a numerical simulation model of multi-level PRB treatment of acid mine water in coal mines is constructed; the numerical simulation model is used to simulate the migration and diffusion of characteristic pollutants of acid mine water in coal mines in PRB fillers at each level.

3. The optimization method for multi-stage PRB in-situ remediation of acid mine water in coal mines according to claim 2, characterized in that: The size and hydrogeological parameters of each level of PRB filler in the two-dimensional geometric model are assigned to construct a numerical simulation model for treating acid mine water with multi-level PRB, specifically including: Using limestone, coconut shell biochar, and anion exchange resin D201 as multi-stage PRB fillers, the hydrogeological parameters of each stage of PRB fillers are obtained; the hydrogeological parameters include filler porosity, permeability, dispersivity, and chemical reaction parameters; In the numerical simulation software, the porosity, permeability, dispersivity and chemical reaction parameters of each level of PRB filler are assigned, and the groundwater seepage model and reaction migration model of acid mine water in each filling material of multi-level PRB are constructed. The migration and diffusion of characteristic pollutants of acid mine water in coal mines in each level of filler are simulated to obtain a numerical simulation model.

4. The optimization method for multi-stage PRB in-situ remediation of acid mine water in coal mines according to claim 3, characterized in that: The numerical simulation model includes a groundwater seepage model and a reaction migration model; the groundwater seepage model has the formula: ； in, K is the permeability coefficient of the aquifer, H is the water level of the aquifer, S s is the water storage rate of the porous medium, t For time, x and y All are independent variables; The reaction migration model has the following formula: ； in, c is the solute concentration, D L is the hydrodynamic diffusion tensor, u is the actual groundwater flow velocity, c p is the concentration of pollutant adsorbed per unit solid dry weight, ρ b is the dry density of the solid medium, ρ b c p is the mass of solute adsorbed on the porous medium; k b c p is the chemical reaction mass of the solute in the porous medium.

5. The optimization method for multi-stage PRB in-situ remediation of acid mine water in coal mines according to claim 1, characterized in that: The objective function is the service life and pollutant treatment capacity of the multi-stage PRB, and the hydrogeological parameters of the multi-stage PRB filler and the concentration load of characteristic pollutants are the constraints, specifically including: To maximize the effective service life of multi-stage PRB f 1 , maximize the effective water treatment volume of multi-stage PRB f 2 , maximize the amount of Fe treated by multi-stage PRB f 3 , Maximizing multi-stage PRB treatment of SO4 2- Amount f 4 As the objective function, the length of limestone L PRB1 , biochar length L PRB2 , anion exchange resin length L PRB3 , initial water head H in , limestone porosity n PRB1 , anion exchange resin porosity n PRB2 , biochar porosity n PRB3 , Limestone Permeability K PRB1 , anion exchange resin permeability coefficient K PRB2 , Biochar permeability coefficient K PRB3 , Dispersion α , Influent Fe concentration c Fe 、Influent SO4 2- concentration c SO4 2- As the constraint condition, the calculation formula is as follows: Max f 1 = min( D Fe , D SO4 2- ); Max f 2 = v × f 1 × A ; Max f 3 = f 2 × c Fe ; Max f 4 = f 2 × c SO4 2- ; s.t.{∑ L PRBi = 6; L PRBi 、 n PRBi 、 K PRBi 、 H in 、 c Fe ∈0.01×N; α 、 c SO4 2- ∈0.1×N, i = 1,2,3}; in, D Fe and D SO4 2- For the i The time taken for a characteristic pollutant from the start of the simulation to the time when the concentration at the outlet exceeds the limit of the groundwater quality standard. v is the Darcy velocity at the outlet, A is the cross-sectional area perpendicular to the water flow direction, c Fe is the influent Fe concentration, c SO4 2- For influent SO4 2- concentration, N is the reference value set.

6. The optimization method for multi-stage PRB in-situ remediation of acid mine water in coal mines according to claim 1, characterized in that: The prediction effect is evaluated by the correlation coefficient between each objective function value and the prediction result, specifically including: Construct a back propagation neural network model in Matlab and divide the samples into training set and test set according to the set ratio; The training set is input into the back propagation neural network model for training, and the test set is input into the trained back propagation neural network model for prediction; Calculate the determination coefficient R between the prediction results and the numerical simulation results 2 , root mean square error RMSE, mean absolute error MAE and mean bias error MBE to evaluate the prediction effect of the back propagation neural network model; The back propagation neural network model with the best prediction effect is saved and used as an alternative model for the numerical simulation model.

7. The optimization method for multi-stage PRB in-situ remediation of acid mine water in coal mines according to claim 1, characterized in that: The coupling of the surrogate model of the numerical simulation model with the fast non-dominated sorting genetic algorithm specifically includes: The Latin hypercube sampling method LHS in Matlab software was used to generate the initial population; The initial population is input into the surrogate model of the numerical simulation model for prediction, and the corresponding objective function value is output. The Pareto frontier of the objective function value is calculated by a fast non-dominated sorting genetic algorithm, and the Pareto optimal solution set of multi-stage PRB for treating acid mine water in coal mines is output.

Images