Multi-target well group precipitation optimization calculation method based on intelligent algorithm

Abstract

The invention discloses a calculation method for multi-target well group precipitation optimization based on an intelligent algorithm. The calculation method comprises the following steps: constructing a foundation pit precipitation seepage-settlement mathematical model in a research area; performing parameter assignment and parameter correction on the model; establishing a target function; establishing a judgment matrix; determining each sub-target weight of the target function according to the judgment matrix to obtain a final target function, forming an optimization solution algorithm by the final target function and the genetic algorithm, and coupling the optimization solution algorithm with the established MODFLOW underground water foundation pit rainfall seepage-settlement mathematical model through Flopy to obtain a target function model. And the optimization solution algorithm continuously reads MODFLOW simulation results and performs iterative calculation until an optimization target is achieved. According to the method, the functions of opening and closing of the well, the single well water pumping amount and the like can be automatically optimized, a corresponding solution set is given, the influence of foundation pit dewatering construction on the environment is effectively reduced, the construction cost is reduced, and a new technical means is provided for modern foundation pit well group dewatering design.

Description

Technical Field

[0001] The invention relates to a multi-objective well group dewatering optimization calculation method based on an intelligent algorithm, and belongs to the field of geological environment protection and well group dewatering optimization design calculation. Background Art

[0002] Dewatering construction is an important topic in the construction process of various large-scale projects. Understanding the hydrogeological conditions of the construction site and formulating a reasonable well group dewatering plan in advance are the prerequisites for avoiding secondary disasters caused by dewatering construction and ensuring the safety and economy of engineering construction. With the continuous development of computers, numerical simulation methods have been widely used in the simulation and solution of various engineering problems. Under the conditions of complex boundary conditions and formation conditions, the numerical method can obtain a good approximate solution, and the visualization platform can help customers intuitively understand the site conditions. However, in the current dewatering design scheme, most of them adopt the position of equidistant wells, and the total water inflow is divided by the water inflow of a single well to obtain the criterion of the number of designed wells. Although this method is simple and easy, it is difficult to apply to complex geological environments. Setting "excess" pumping volume increases construction costs and may also cause various secondary geological disasters. Therefore, it is necessary to optimize the design of the dewatering well group in advance to determine the reasonable number of wells and drainage volume, ensure the safe construction of the foundation pit, reduce the impact on the surrounding environment, and achieve the best dewatering effect.

[0003] In the optimization design of foundation pit dewatering, the problems faced are often composed of multiple conflicting and influencing objectives. For example, if the number of pumping wells is increased, the greater the pumping volume, the safer the underground structure will be. However, at the same time, the comprehensive dewatering cost will also increase. Excessive pumping volume will also cause ground subsidence, which will affect the safety of construction. This requires engineering decision-makers to comprehensively consider technical feasibility, economic benefits and environmental impact, and to find the optimal solution to the problem to be optimized as much as possible by adjusting the weight of each objective. With the rapid development of intelligent algorithms, the optimal design solution can be quickly and intelligently selected by combining numerical simulation and optimization algorithms.

[0004] Well group dewatering methods can be generally divided into objective function method, large well method and finite element method. The objective function method finds the optimal design parameters that meet the constraints by taking the minimum or maximum value of a specific target. However, in existing research, the optimization research and application of the layout scheme of pumping well groups are very limited. Researchers usually use the gradient descent method to solve the objective function problem, and the application of intelligent optimization methods in solving the objective function is less. Although MODFLOW has become the most mature program in groundwater simulation, its coupling optimization design combined with intelligent optimization algorithms needs to be further improved. Therefore, the present invention constructs a multi-objective well group dewatering optimization model based on GA and FloPy, so that the optimization algorithm and the numerical simulation model can be tightly coupled, and a calculation method that can realize the automatic switching of pumping wells, the automatic optimization of single well water inflow and the location of the well is proposed. The optimization calculation method has good effectiveness and reliability, and can provide a new technical means for the optimization design of well group dewatering. Summary of the invention

[0005] The present invention discloses a multi-objective well group dewatering optimization calculation method based on an intelligent algorithm. In the optimization design of foundation pit dewatering, a multi-objective well group dewatering optimization model based on GA and FloPy is constructed, so that the optimization algorithm and the numerical simulation model can be closely coupled. At the same time, the weight of each objective in the function is determined by the hierarchical analysis method, and an optimization calculation method that can realize the automatic switching of the pumping well, the automatic optimization of the single well water inflow and the location of the well is realized. Based on a rigorous mathematical model and a mature intelligent algorithm, the present invention avoids the problem of conservative design of conventional dewatering schemes and large total water inflow of the vertical shaft. It can also improve the economic benefits of the project, save groundwater resources, and provide a new technical means for the optimization design of well group dewatering.

[0006] In order to solve the above technical problems, the technical solutions adopted by the present invention are as follows:

[0007] A calculation method for multi-objective well group dewatering optimization based on intelligent algorithm includes the following steps:

[0008] (1) Based on the on-site topography, stratum lithology, hydrogeological conditions, and engineering geological conditions, the scope of the study area is comprehensively determined, the aquifer structure, boundary conditions, and initial conditions around the foundation pit are determined, and a mathematical model of foundation pit dewatering seepage-sedimentation in the study area is constructed;

[0009] The mathematical model of foundation pit dewatering seepage-settlement is established as follows:

[0010] ① Mathematical model of groundwater flow

[0011]

[0012] Where: W is the sink-source function, m / d; when the groundwater type is confined water, F = KM and E = μ* , where K is the permeability coefficient of the aquifer, m / d; M is the thickness of the aquifer, m; μ * is the elastic water storage coefficient; when the groundwater type is diving, F=K(HZ) and E= μ , H is the water table height, m; Z is the aquifer impermeable bottom plate height, m; μ is the gravity water supply degree; t is the simulation time, t = 0 represents the initial time; Ω 1 is the first kind of boundary; Ω 2 is the second type of boundary; n is the boundary Ω 2 Normal direction; K is the boundary Ω 2 The permeability coefficient in the normal direction, m / d; q(x, y, z, t) is the flow rate per unit area of ​​the boundary, m 3 / (d·m 2 ), q = 0 indicates the zero flow boundary, which is the water-blocking boundary in this study; H 0 (x, y, z) is the initial water level of the aquifer, m; H 1 (x, y, z, t) is the boundary Ω 1 Known functions on ;

[0013] ② Stratum subsidence model

[0014] Based on Terzaghi's effective stress principle, only the vertical stress of the stratum is considered, and the horizontal stress is ignored. The total stress on any plane in the soil is composed of effective stress and pore water pressure. The vertical settlement model is:

[0015]

[0016] Where: Δb is the settlement, mm; Ω is the seepage area; S sf is the skeleton water storage rate, m -1 , when the water level in the stratum is lower than the previous lowest water level, S sf is the water storage rate of the inelastic skeleton (S sfv ), when the water level is higher than the previous minimum water level, S sf is the elastic skeleton water storage rate (S sfe ) ; Δh is the change of water head, m; b is the thickness of the corresponding formation, m;

[0017] (2) Determine the initial values ​​of the hydrogeological parameters and physical and mechanical parameters of the study area based on the basic data of the study area, preliminary field tests and indoor tests, and assign parameters to the mathematical model of foundation pit dewatering and seepage-sedimentation in the study area;

[0018] (3) running the foundation pit dewatering seepage-sedimentation mathematical model obtained after the assignment in step (2), and performing parameter correction based on the measured data of the water level observation points to ensure that the simulated groundwater seepage field model is relatively consistent with the natural situation;

[0019] (4) Establishing the objective function: The solution of the well group dewatering optimization model can be transformed into an optimization problem. The economic cost of well completion, operating economic cost, water level drop at the observation point and environmental cost of the dewatering scheme are used as sub-objectives to form the objective function. The minimum total value of the objective function is used as the overall optimization goal.

[0020] (5) Establishing a judgment matrix: Since there are multiple sub-goals in the objective function established in step (4), these sub-goals of the precipitation optimization model often affect each other. Different weights can be multiplied according to the importance of each sub-goal;

[0021] (6) Model coupling and optimization solution: According to the judgment matrix in step (5), the weights of each sub-objective of the objective function in step (4) are confirmed to obtain the final objective function. The final objective function and the genetic algorithm together constitute the optimization solution algorithm. The optimization solution algorithm is coupled with the MODFLOW groundwater foundation pit dewatering seepage-sedimentation mathematical model established in step (3) through Flopy. The optimization solution algorithm continuously reads the MODFLOW simulation results (such as the settlement and water head at the observation point) and performs iterative calculations until the optimization target is achieved.

[0022] The aquifer structure of the model in the above step (1) includes a phreatic aquifer and a confined aquifer, the boundary conditions include a first-type constant head boundary, a second-type constant flow boundary and a third-type mixed boundary, and the initial conditions include an initial head condition and an initial stress condition.

[0023] In the above step (2), the initial values ​​of the hydrogeological parameters and physical and mechanical parameters required for the calculation are obtained based on the on-site investigation and research report, on-site pumping test, micro-water test, water injection test, indoor constant head test, variable head test and geotechnical test; the hydrogeological parameters include permeability, porosity, volume compression coefficient and specific gravity, etc.

[0024] In the above step (3), the means of parameter correction include but are not limited to: setting up simulated observation wells and actual observation well data for fitting and parameter adjustment and using PEST. module to invert parameters.

[0025] In the above step (4), the objective function of the well group dewatering optimization design contains two types of variables: decision variables and state variables; the decision variables are the objects to be optimized, including the number, location and water yield of the pumping wells, and the best dewatering effect can be achieved by adjusting their combination; due to the environmental restrictions of the project site, the location of the pumping wells cannot be set arbitrarily; therefore, before the optimization model is run, it is necessary to select the possible locations of the pumping wells in advance and determine the maximum number of pumping wells; the optimization algorithm will search among the possible locations of the pumping wells, and then determine the best layout plan for the pumping well group. This process is completed by defining the opening or closing of the pumping wells as binary variables. , some decision variables in the objective function can only be selected from 0 or 1, which is called 0-1 integer programming. The 0-1 integer programming in the genetic algorithm used this time represents the switch status of the pumping well. If the pumping well is coded as 1 in the algorithm, it means that the pumping well is working, otherwise it means that the pumping well has stopped pumping water; the groundwater level of the aquifer and the maximum settlement limit of the observation point are used as state variables of the optimization model. They are the output results of the MODFLOW model after the decision variables are input and run, and they change with the change of the decision variables; the state variables and decision variables serve as a bridge between the MODFLOW program and the genetic algorithm program in the precipitation optimization model;

[0026] The objective function is composed of the economic cost of well completion, operation economic cost, water level drop at the observation point and environmental cost as sub-goals, and the minimum total value of the objective function is taken as the overall optimization goal. Since there are multiple sub-goals in the objective function, it is necessary to solve it by assigning reasonable weight values ​​to each sub-goal. The established objective function formula is as follows:

[0027]

[0028] The constraints corresponding to the objective function are:

[0029] n z ≥n w (4)

[0030] Q max ≥Q i ≥Q min (5)

[0031] In the formula, Y is the objective function value, α 1 is the weight coefficient of the economic cost of well completion, c 1 The cost of drilling and cementing the pumping well; 2 is the weight coefficient of the economic cost of pumping water, c 2 is the cost per unit of water pumping; X i is a binary variable, X i =1 means the pumping well works normally, X i =0 means the pumping well is not working; Qi is the water yield of the normal pumping well, n obs is the number of observation points, β 1 is the target water level weight coefficient of the observation point, β 2 is the weight coefficient of the maximum settlement limit of the observation point, h i is the water level value at the observation point, H i is the target water level value, S i is the settlement of the observation point, n z is the maximum number of pumping wells allowed, n w is the number of pumping wells actually in operation, Q min , Q max They are the minimum and maximum water yields of a single well, respectively.

[0032] In the above step (5), the analytic hierarchy process (AHP) is used to analyze the weights of each sub-goal. AHP can not only transform the decision maker's judgment into a comparison of the importance of several factors, thereby transforming the qualitative judgment that is difficult to quantify into an operational comparison of importance; AHP can also reflect the basic characteristics of decision-making thinking, that is, based on the observation of the research target, data collection, statistical analysis and summary, it finally obtains a quantitative weight result evaluation, overcoming the shortcomings of other analysis methods that avoid the subjective judgment of decision makers.

[0033] In the above step (5), the four sub-goals in the objective function: economic cost of well completion, economic cost of operation, target water level value and maximum settlement limit are used as the elements of the indicator layer, and the weight analysis of each sub-goal is used as the target layer, so as to construct a hierarchical structure model of sub-goal weight analysis in the objective function.

[0034] In the above step (5), define the n-order judgment matrix a ij , compare the importance of each sub-goal in the indicator layer, and use the mutual ratios between the indicators as elements in the matrix. The importance scale is quantified in the form of expert scoring. The judgment matrix scale is shown in Table 1. At this time, the judgment matrix A is obtained:

[0035]

[0036] Table 1 Judgment matrix scale

[0037]

[0038] In the above step (5), after the judgment matrix A is constructed, the maximum eigenvalue and eigenvector of A need to be solved and tested according to the consistency test index:

[0039]

[0040] Where CI is the consistency test index of matrix A;max is the maximum eigenvalue of matrix A; n is the order of matrix A; CR is the consistency ratio, when CR < 0.1, the matrix is ​​judged to be consistent; RI is the n-order average random consistency index. RI is the n-order average random consistency index. The corresponding relationship between the two can be found from the average random consistency test index table, as shown in Table 2;

[0041] Table 2 Average random consistency test index table

[0042]

[0043] In the above step (6), FloPy is used to couple the foundation pit dewatering seepage-sedimentation mathematical model established by MODFLOW with the genetic algorithm for solution. The genetic algorithm gradually adjusts and evolves the population toward the direction containing the optimal solution by selecting, crossing and mutating the initial population. In practical application, the genetic algorithm can obtain a definite global optimal solution according to the objective function; the well switch is defined as a binary problem, and the decision variable decision can only be selected between 0 or 1. Compared with other algorithms, the genetic algorithm can be well compatible with this optimization requirement. When using the model, FloPy is used to create a MODFLOW groundwater model based on the hydrogeological model, and then the genetic algorithm is used to solve the objective function established based on FloPy. The objective function value will read the MODFLOW simulation results (such as the settlement and head of the observation point) for iterative calculation.

[0044] The technologies not mentioned in the present invention are all referred to the prior art.

[0045] The invention discloses a calculation method for optimizing well group dewatering based on a multi-objective intelligent algorithm. The method establishes an accurate three-dimensional groundwater seepage model of a study area by using a MODFLOW module, and then constructs an objective function with sub-objectives such as well construction cost, operation cost, water level drop at an observation point and environmental cost. The AHP method is used to assign weights according to the importance of each sub-objective and the degree of influence on the project, and a multi-objective well group dewatering optimization model based on GA and FloPy is constructed. This allows the optimization algorithm to be tightly coupled with a numerical simulation model, and proposes a calculation method that can realize automatic switching of pumping wells, automatic optimization of single well water inflow and well location. The optimization calculation method has good effectiveness and reliability, effectively reduces the impact of foundation pit dewatering construction on the environment, reduces construction costs, and provides a new technical means for optimizing the design of well group dewatering. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic diagram of a flow chart of a multi-objective well group dewatering optimization calculation method based on an intelligent algorithm according to an embodiment of the present invention;

[0047] Figure 2 This is a diagram showing the setting of a well group dewatering optimization verification model in an embodiment of the present invention;

[0048] Figure 3 It is a cloud map of water level and settlement of the site under the multi-objective well group dewatering optimization calculation based on intelligent algorithm in the embodiment of the present invention;

[0049] Figure 4 The water level and settlement cloud map of the site calculated by the "big well method" in the embodiment of the present invention;

[0050] Figure 5 This is a comparison chart of various indicators calculated by different methods in the embodiments of the present invention; DETAILED DESCRIPTION

[0051] In order to better understand the present invention, the content of the present invention is further explained below in conjunction with the embodiments, but the content of the present invention is not limited to the following embodiments.

[0052] A multi-objective well group dewatering optimization calculation method based on intelligent algorithm, firstly, uses MODFLOW module to establish an accurate three-dimensional groundwater seepage model of the study area. The objective function is constructed with sub-objectives such as well completion economic cost, operation economic cost, water level drop at observation point and environmental cost, and the AHP method is used to assign weights according to the importance of each sub-objective and the degree of influence on the project. Finally, Flopy is used to couple the MODFLOW model with GA, and the solution is obtained through the good compatibility and optimization of genetic algorithm to obtain the best well group dewatering scheme. The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0053] Example 1: This example provides a multi-objective well group dewatering optimization calculation method based on intelligent algorithm. The calculation process is shown as follows: Figure 1 As shown, it is achieved by the following steps.

[0054] (1) In order to verify the effectiveness and superiority of the multi-objective well group dewatering optimization calculation method based on intelligent algorithm, an ideal well group dewatering optimization verification model was designed to simulate the actual situation of foundation pit dewatering projects. The model is a single-layer phreatic aquifer with a thickness of 80m. The model extends 105m in the X and Y directions and is divided into 441 units by a grid with a side length of 5m. The model boundaries are all set as fixed head boundaries. The fixed head and initial head are both set to 80m. The bottom is set as a water-proof boundary. The site is divided into three areas with different hydrogeological parameters. A square foundation pit with a side length of 25m is set in the site. Eight alternative pumping wells are arranged around the foundation pit. All pumping wells are submerged complete wells. The center point of the foundation pit is used as a water level observation point, and a settlement observation point is arranged on the east side of the model to simulate the settlement of important locations caused by foundation pit dewatering. The dewatering target of this time is: within 10 days, the water level at the center point of the foundation pit drops by 3m, and the settlement value of the settlement observation point is less than 5mm. The specific settings of the model are as follows: Figure 2 shown.

[0055] (2) Establishment of mathematical model of foundation pit dewatering seepage-settlement

[0056] ① Mathematical model of groundwater flow

[0057] The model groundwater type is phreatic water, and its governing equation is

[0058]

[0059] Where: D is the seepage area; H is the water level, m; K x , K y , K z are the main permeability coefficients in the x, y, and z directions, m / d; h is the water table height, m; ε is the sink-source function, m / d; μd is the water supply degree; t is the simulation time, t = 0 represents the initial time; Ω 1 is the first kind of boundary; Ω 2 is the second type of boundary; n is the boundary Ω 2 Normal direction; K is the boundary Ω 2 The permeability coefficient in the normal direction, m / d; q(x, y, z, t) is the flow rate per unit area of ​​the boundary, m 3 / (d·m 2 ), q = 0 indicates the zero flow boundary, which is the water-blocking boundary in this study; H 0 (x, y, z) is the initial water level of the aquifer, m; H 1 (x, y, z, t) is the boundary Ω 1 Known functions on .

[0060] ② Formation settlement model. Based on Terzaghi's effective stress principle, only the vertical stress of the formation is considered, and the horizontal stress is ignored. The total stress on any plane in the soil is composed of effective stress and pore water pressure. The vertical settlement model is:

[0061]

[0062] Where: Δb is the settlement, mm; Ω is the seepage area; S sf is the skeleton water storage rate, m -1 , when the water level in the stratum is lower than the previous lowest water level, S sf is the water storage rate of the inelastic skeleton (S sfv ), when the water level is higher than the previous minimum water level, S sf is the elastic skeleton water storage rate (S sfe ), Δh is the change of water head, m, and b is the corresponding formation thickness, m.

[0063] (3) Based on the on-site investigation and research report, on-site pumping test, micro-water test, water injection test, indoor constant head test, variable head test and geotechnical test, the initial values ​​of the hydrogeological parameters and physical and mechanical parameters required for the calculation are obtained, and the parameters of the mathematical model of foundation pit dewatering and seepage-sedimentation in the study area are assigned; among them, the hydrogeological parameters include permeability, porosity, volume compression coefficient and specific gravity.

[0064] (4) The mathematical model of foundation pit dewatering seepage and settlement obtained in step (3) is run, and parameters are corrected through measured data from water level observation points to ensure that the simulated groundwater seepage field model is consistent with the natural conditions; wherein, the means of parameter correction include: setting simulated observation wells and measured data from field observation wells for fitting and parameter adjustment and using the PEST. module to invert parameters.

[0065] (5) Establishing the objective function: The solution of the well group dewatering optimization model can be transformed into an optimization problem. The economic cost of well completion, operating economic cost, water level drop at the observation point and environmental cost of the dewatering scheme are used as sub-objectives to form the objective function. The minimum total value of the objective function is used as the overall optimization goal.

[0066] The objective function of well group dewatering optimization design contains two types of variables: decision variables and state variables; decision variables are the objects to be optimized, including the number, location and water yield of pumping wells, and the best dewatering effect can be achieved by adjusting their combination; due to the environmental restrictions of the project site, the location of the pumping wells cannot be set arbitrarily; therefore, before the optimization model is run, it is necessary to select the possible locations of the pumping wells in advance and determine the maximum number of pumping wells; the optimization algorithm will search among the possible locations of the pumping wells, and then determine the best layout of the pumping well group. This process is completed by defining the opening or closing of the pumping wells as binary variables. The objective function Some decision variables in the algorithm can only be selected between 0 or 1, which is called 0-1 integer programming. The 0-1 integer programming in the genetic algorithm used this time represents the switch status of the pumping well. If the pumping well is coded as 1 in the algorithm, it means that the pumping well is working, otherwise it means that the pumping well has stopped pumping water; the groundwater level of the aquifer and the maximum settlement limit of the observation point are used as state variables of the optimization model. They are the output results of the MODFLOW model after the decision variables are input and run, and they change with the change of the decision variables; the state variables and decision variables serve as a bridge between the MODFLOW program and the genetic algorithm program in the precipitation optimization model;

[0067] The objective function is composed of the economic cost of well completion, operation economic cost, water level drop at the observation point and environmental cost as sub-goals, and the minimum total value of the objective function is taken as the overall optimization goal. Since there are multiple sub-goals in the objective function, it is necessary to solve it by assigning reasonable weight values ​​to each sub-goal. The established objective function formula is as follows:

[0068]

[0069] The constraints corresponding to the objective function are:

[0070] n z ≥n w (4)

[0071] Q max ≥Q i ≥Q min (5)

[0072] In the formula, Y is the objective function value, α 1 is the weight coefficient of the economic cost of well completion, c 1 The cost of drilling and cementing the pumping well; 2 is the weight coefficient of the economic cost of pumping water, c 2 is the cost per unit of water pumping; X i is a binary variable, X i =1 means the pumping well works normally, X i =0 means the pumping well is not working; Q i is the water yield of the normal pumping well, n obs is the number of observation points, β 1 is the target water level weight coefficient of the observation point, β 2 is the weight coefficient of the maximum settlement limit of the observation point, h i is the water level value at the observation point, H i is the target water level value, S i is the settlement of the observation point, n z is the maximum number of pumping wells allowed, n w is the number of pumping wells actually in operation, Q min , Q max They are the minimum and maximum water yields of a single well, respectively.

[0073] (6) Establishing a judgment matrix: Since there are multiple sub-goals in the objective function established in step (5), these sub-goals of the precipitation optimization model often affect each other. Different weights can be multiplied according to the importance of each sub-goal;

[0074] The analytic hierarchy process (AHP) is used to analyze the weight of each sub-goal. AHP can not only transform the decision maker's judgment into a comparison of the importance of several factors, thereby transforming the qualitative judgment that is difficult to quantify into an operational comparison of importance; AHP can also reflect the basic characteristics of decision-making thinking, that is, based on the observation of research objectives, data collection, statistical analysis and summary, and finally obtain a quantitative weight result evaluation, which overcomes the shortcomings of other analysis methods that avoid the subjective judgment of decision makers.

[0075] The four sub-goals in the objective function: economic cost of well completion, economic cost of operation, target water level value and maximum settlement limit are taken as the elements of the indicator layer, and the weight analysis of each sub-goal is taken as the target layer to construct a hierarchical structure model of sub-goal weight analysis in the objective function.

[0076] Define the n-order judgment matrix a ij , compare the importance of each sub-goal in the indicator layer, and use the mutual ratios between the indicators as elements in the matrix. The importance scale is quantified in the form of expert scoring. The judgment matrix scale is shown in Table 1. At this time, the judgment matrix A is obtained:

[0077]

[0078] Table 1 Judgment matrix scale

[0079]

[0080] After the judgment matrix A is constructed, the maximum eigenvalue and eigenvector of the solution A are required to be tested according to the consistency test index:

[0081]

[0082] Where CI is the consistency test index of matrix A; max is the maximum eigenvalue of matrix A; n is the order of matrix A; CR is the consistency ratio. When CR<0.1, the matrix is ​​judged to be consistent; R is the n-order average random consistency index. The corresponding relationship between the two can be found from the average random consistency test index table, as shown in Table 2. When n=4, RI=0.89.

[0083] Table 2 Average random consistency test index table

[0084]

[0085] (7) Model coupling and optimization solution: According to the judgment matrix in step (6), the weights of each sub-objective of the objective function in step (5) are confirmed to obtain the final objective function. The final objective function and the genetic algorithm together constitute the optimization solution algorithm. The optimization solution algorithm is coupled with the MODFLOW groundwater foundation pit dewatering seepage-sedimentation mathematical model established in step (4) through Flopy. The optimization solution algorithm continuously reads the MODFLOW simulation results (such as the settlement and water head at the observation point) and performs iterative calculations until the optimization target is achieved.

[0086] In this example, the coupling model is called to calculate the verification model. In this verification, the weight coefficients in the objective function are assigned the same value. The proposed optimization model will automatically optimize the pumping parameters of the well group according to the constraints. If parallel tests are carried out under the same conditions, the optimization model will give the corresponding series of solutions. The verification model is calculated 10 times in parallel to ensure that the water level at the center of the foundation pit drops by 3m, and the impact of pumping on the settlement observation point is minimized as much as possible, while reducing the comprehensive cost of precipitation, including construction costs and operating costs.

[0087] Comparison of calculations using the traditional large well method to arrange a group of precipitation wells

[0088] At present, the "big well method" is mainly used in engineering to calculate the total water inflow of the foundation pit, and then the total water inflow is evenly distributed to each pumping well to determine the foundation pit dewatering well group. The advantages of this method are simple calculation and easy operation. It is calculated according to the complete well formula of the phreatic aquifer specified in the specification:

[0089]

[0090] Where: Q is the water inflow from the foundation pit (m 3 ), K is the permeability coefficient of the aquifer (m / d), S is the water level drawdown (m), H is the thickness of the aquifer (m), r w is the equivalent radius (m); when the foundation pit shape is rectangular, a, b are the length and width of the rectangle respectively (m).

[0091] The results of the large well method and ten parallel optimization calculations are shown in Table 3. The site water level and ground subsidence cloud diagrams are shown in Table 3. Figure 3 The total water inflow of the foundation pit of the verification model was calculated using the “big well method”. 3 ) are evenly distributed to 8 standby pumping wells. After 10 days of well group pumping work, the site water level and ground subsidence cloud map is as follows Figure 4 As shown in the figure, the well construction cost is 500 yuan / well and the pumping operation fee is 1 yuan / m 3 , the comprehensive calculation results are compared with Figure 5 shown.

[0092] Table 3 Results of 10 parallel optimizations

[0093]

[0094] The multi-objective well group dewatering optimization calculation method based on intelligent algorithms has shown good calculation results. Compared with the calculation results of the "big well method", the target water level obtained by the water level observation point is more accurate, and the settlement of a certain point on the site can be precisely controlled. The optimization algorithm will give a solution set under the restricted conditions. Overall, the optimization algorithm reduces the pumping volume by 20.3%, the settlement by 53.7%, and the construction cost by 45.9% compared with the calculation results of the big well method. The optimization algorithm is more accurate, safe and economical to optimize the calculation of the foundation pit pumping well group.

[0095] The multi-objective well group dewatering optimization calculation method based on intelligent algorithm can realize the automatic optimization of well opening and closing, single well pumping volume and other functions, and give the corresponding solution set, which can effectively reduce the impact of foundation pit dewatering construction on the environment and reduce construction costs, providing a new technical means for the modern foundation pit well group dewatering design.

Claims

1. A calculation method for multi-objective well group dewatering optimization based on intelligent algorithm, characterized by: The steps include: (1) Based on the on-site topography, stratum lithology, hydrogeological conditions and engineering geological conditions, the scope of the study area is comprehensively determined, the aquifer structure, boundary conditions and initial conditions around the foundation pit are determined, and a mathematical model of foundation pit dewatering seepage-sedimentation in the study area is constructed; The mathematical model of foundation pit dewatering seepage-settlement is established as follows: ① Mathematical model of groundwater flow Where: W is the sink-source function, m / d; when the groundwater type is confined water, F = KM and E = μ * , where K is the permeability coefficient of the aquifer, m / d; M is the thickness of the aquifer, m; μ * is the elastic water storage coefficient; when the groundwater type is diving, F = K (HZ) and E = μ, H is the diving surface height, m; Z is the height of the aquifer impermeable bottom plate, m; μ is the gravity water supply degree; t is the simulation time, t = 0 represents the initial time; Ω1 is the first type of boundary; Ω2 is the second type of boundary; n is the normal direction of the boundary Ω2; K is the permeability coefficient in the normal direction of the boundary Ω2, m / d; q (x, y, z, t) is the flow rate per unit area section of the boundary, m 3 / (d·m 2 ), q = 0 represents the zero flow boundary, which is the impermeable boundary in this study; H0 (x, y, z) is the initial water level of the aquifer, m; H1 (x, y, z, t) is the known function on the boundary Ω1; ② Stratum subsidence model Based on Terzaghi's effective stress principle, only the vertical stress of the stratum is considered, and the horizontal stress is ignored. The total stress on any plane in the soil is composed of effective stress and pore water pressure. The vertical settlement model is: Where: Δb is the settlement, mm; Ω is the seepage area; S sf is the skeleton water storage rate, m -1 , when the water level in the stratum is lower than the previous lowest water level, S sf is the water storage rate of the inelastic skeleton (S sfv ), when the water level is higher than the previous minimum water level, S sf is the elastic skeleton water storage rate (S sfe ) ; Δh is the change of water head, m; b is the thickness of the corresponding formation, m; (2) Determine the initial values ​​of the hydrogeological parameters and physical and mechanical parameters of the study area based on the basic data of the study area, preliminary field tests and indoor tests, and assign parameters to the mathematical model of foundation pit dewatering and seepage-sedimentation in the study area; (3) running the foundation pit dewatering seepage-sedimentation mathematical model obtained after the assignment in step (2), and performing parameter correction based on the measured data of the water level observation points to ensure that the simulated groundwater seepage field model is relatively consistent with the natural situation; (4) The economic cost of well completion, operation economic cost, water level drop at the observation point and environmental cost of the dewatering scheme are used as sub-objectives to form an objective function, and the minimum total value of the objective function is used as the overall optimization goal; (5) According to the importance of each sub-goal, each sub-goal is multiplied by a different weight to form a judgment matrix; (6) According to the judgment matrix in step (5), the weights of each sub-objective of the objective function in step (4) are confirmed to obtain the final objective function. The final objective function and the genetic algorithm together constitute an optimization solution algorithm. The optimization solution algorithm is coupled with the MODFLOW groundwater foundation pit dewatering seepage-sedimentation mathematical model established in step (3) through Flopy. The optimization solution algorithm continuously reads the MODFLOW simulation results and performs iterative calculations until the optimization goal is achieved.

2. The calculation method for multi-objective well group dewatering optimization based on intelligent algorithm according to claim 1 is characterized by: In step (2), the initial values ​​of the hydrogeological parameters and physical and mechanical parameters required for the calculation are obtained based on the on-site investigation and research report, on-site pumping test, micro-water test, water injection test, indoor constant head test, variable head test and geotechnical test; the hydrogeological parameters include permeability, porosity, volume compression coefficient and specific gravity.

3. The calculation method for multi-objective well group dewatering optimization based on intelligent algorithm according to claim 1 or 2, characterized in that: In step (3), the means of parameter correction include but are not limited to: setting simulated observation wells and field observation well measured data for fitting and parameter adjustment and using PEST. module to invert parameters.

4. The calculation method for multi-objective well group dewatering optimization based on intelligent algorithm according to claim 1 or 2, characterized in that: In step (4), the objective function contains two types of variables: decision variables and state variables; The decision variables are the objects to be optimized, including the number, location and water yield of the pumping wells. The best precipitation effect can be achieved by adjusting their combination. Due to the environmental restrictions of the project site, the location of the pumping wells cannot be set arbitrarily. Therefore, before the optimization model is run, it is necessary to select the possible locations of the pumping wells and determine the maximum number of pumping wells in advance. The optimization algorithm will search among the possible locations of the pumping wells, and then determine the best layout of the pumping well group. This process is completed by defining the opening or closing of the pumping wells as binary variables. Some decision variables in the objective function can only be set at 0 or 1, which is called 0-1 integer programming. The 0-1 integer programming in the genetic algorithm used this time represents the switch status of the pumping well. If the pumping well is coded as 1 in the algorithm, it means that the pumping well is working, otherwise it means that the pumping well has stopped pumping water; the groundwater level of the aquifer and the maximum settlement limit of the observation point are used as state variables of the optimization model. They are the output results of the MODFLOW model after the decision variables are input and run, and they change with the change of the decision variables; the state variables and decision variables serve as a bridge between the MODFLOW program and the genetic algorithm program in the precipitation optimization model; The objective function is composed of the economic cost of well completion, operation economic cost, water level drop at the observation point and environmental cost as sub-goals, and the minimum total value of the objective function is taken as the overall optimization goal. Since there are multiple sub-goals in the objective function, it is necessary to solve it by assigning reasonable weight values ​​to each sub-goal. The established objective function formula is as follows: The constraints corresponding to the objective function are: n z ≥n w (4) Q max ≥Q i ≥Q min (5) Where Y is the objective function value, α1 is the weight coefficient of the economic cost of well completion, c1 is the cost of drilling and cementing the pumping well; α2 is the weight coefficient of the economic cost of pumping, c2 is the cost per unit of pumping; X i is a binary variable, X i =1 means the pumping well works normally, X i =0 means the pumping well is not working; Q i is the water yield of the normal pumping well, n obs is the number of observation points, β1 is the weight coefficient of the target water level of the observation point, β2 is the weight coefficient of the maximum settlement limit of the observation point, and h i is the water level value at the observation point, H i is the target water level value, S i is the settlement of the observation point, n z is the maximum number of pumping wells allowed, n w is the number of pumping wells actually in operation, Q min , Q max They are the minimum and maximum water yields of a single well, respectively.

5. The calculation method for multi-objective well group dewatering optimization based on intelligent algorithm according to claim 1 or 2, characterized in that: In step (5), the analytic hierarchy process is used to analyze the weights of each sub-goal. AHP can not only transform the decision maker's judgment into a comparison of the importance of several factors, thereby transforming the difficult-to-quantify qualitative judgment into an operational comparison of importance; AHP can also reflect the basic characteristics of decision-making thinking, that is, based on the observation of the research objectives, data collection, statistical analysis and summary, and finally obtain a quantitative weight result evaluation.

6. The calculation method for multi-objective well group dewatering optimization based on intelligent algorithm according to claim 5 is characterized by: In step (5), the four sub-goals in the objective function: economic cost of well completion, economic cost of operation, target water level value and maximum settlement limit are used as the elements of the indicator layer, and the weight analysis of each sub-goal is used as the target layer to construct a hierarchical structure model of sub-goal weight analysis in the objective function.

7. The calculation method for multi-objective well group dewatering optimization based on intelligent algorithm according to claim 6 is characterized by: In step (5), define the n-order judgment matrix a ij , compare the importance of each sub-goal in the indicator layer, and use the mutual ratios between the indicators as elements in the matrix. The importance scale is quantified in the form of expert scoring. The judgment matrix scale is shown in Table 1. At this time, the judgment matrix A is obtained: Table 1 Judgment matrix scale 8. The calculation method for multi-objective well group dewatering optimization based on intelligent algorithm according to claim 7 is characterized by: In step (5), after the judgment matrix A is constructed, the maximum eigenvalue and eigenvector of the solution A are required to be tested according to the consistency test index: Where, CI is the consistency test index of matrix A; λ max is the maximum eigenvalue of matrix A; n is the order of matrix A; CR is the consistency ratio. When CR<0.1, the matrix is ​​judged to be consistent; RI is the n-order average random consistency index, and RI is the n-order average random consistency index. The corresponding relationship between the two can be found from the average random consistency test index table, as shown in Table 2; Table 2 Average random consistency test index table

Images