Subgrade structure parameter intelligent optimization method and device fusing population inheritance and reverse learning, equipment and storage medium

Abstract

The invention discloses a subgrade structure parameter intelligent optimization method and device fusing population inheritance and reverse learning, equipment and a storage medium, and relates to the technical field of land traffic engineering subgrade structure design, and the method comprises the steps: obtaining a subgrade engineering to-be-optimized parameter set, a subgrade engineering parameter optimization model and a self-adaptive hyper-heuristic algorithm; based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, population inheritance and reverse learning are carried out to generate a population, and target population individual information is determined; based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm and the target population individual information, correlation analysis is carried out to adjust the search direction and the search step length, and target optimization population information is determined; and completing intelligent optimization of roadbed structure parameters based on the target optimization population information. The subgrade parameters are optimized by fusing population heredity and reverse learning, and the optimization efficiency is improved.

Description

Technical Field

[0001] The present application relates to the technical field of subgrade structure design for land transportation engineering, and in particular to a method, device, equipment and storage medium for intelligent optimization of subgrade structure parameters integrating population genetics and reverse learning. Background Art

[0002] In the field of land transportation engineering, the design of roadbed structure is directly related to the safety, stability and economy of the project. With the continuous development of transportation infrastructure construction, the demand for optimization of roadbed structure parameters is increasing to achieve more efficient and economical design solutions. For example, in the construction of transportation infrastructure such as railways and highways, it is necessary to accurately design the parameters of structures such as gravity retaining walls and cantilever retaining walls to ensure their stability and safety under various working conditions, while controlling the project cost.

[0003] At present, the existing practice of subgrade structure parameter design mainly relies on manual experience and trial calculation methods. Based on the experience of previous projects, designers use the combination of retaining structure parameters for trial calculation. If the calculation is passed, further economic comparison will be carried out, otherwise it will not be adopted. In addition, some existing practices have also begun to introduce heuristic algorithms such as genetic algorithms and particle swarm algorithms to realize the automated optimization search process by simulating the survival and reproduction mechanism of biological populations in nature. For example, genetic algorithms evaluate the pros and cons of design schemes through fitness functions, and generate new design schemes through operations such as selection, crossover and mutation; particle swarm algorithms solve problems by simulating the social behavior of bird flocks or fish schools.

[0004] However, the existing manual parameter adjustment is time-consuming and inefficient. It takes a lot of time to try different parameter combinations, and the optimization process is slow. The selection of parameters relies on the experience and expertise of engineers, which has a high threshold. The design preferences of different engineers will cause the solution to be affected by personal subjective factors. In addition, whether it is manual parameter adjustment or classical heuristic algorithms for optimization, it is easy to fall into the dilemma of local optimality and it is difficult to find the global optimal solution. For example, in the complex roadbed structure parameter optimization problem, the algorithm may obtain the optimal solution at a local point, but this solution is not the global optimal solution in the entire problem domain, which limits the optimization effect. Therefore, how to more efficiently and accurately optimize the roadbed structure parameters has become an urgent problem to be solved.

[0005] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0006] The main purpose of this application is to provide a method, device, equipment and storage medium for intelligent optimization of roadbed structure parameters that integrates population genetics and reverse learning, aiming to solve the technical problem of how to intelligently optimize roadbed structure parameters more efficiently and accurately.

[0007] To achieve the above objectives, the present application proposes a method for intelligent optimization of roadbed structure parameters integrating population genetics and reverse learning, wherein the method comprises:

[0008] Obtaining the parameter set to be optimized for the roadbed engineering, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm;

[0009] Based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, population genetics and reverse learning are performed to generate a population, and individual information of the target population is determined;

[0010] Based on the set of parameters to be optimized for the roadbed engineering, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm and the target population individual information, correlation analysis is performed to adjust the search direction and search step, determine the target optimization population information, and complete the intelligent optimization of roadbed structure parameters based on the target optimization population information.

[0011] In one embodiment, the step of obtaining a set of parameters to be optimized for a roadbed engineering project, a roadbed engineering parameter optimization model, and an adaptive hyper-heuristic algorithm comprises:

[0012] Obtaining a combination of roadbed parameters, geotechnical calculation functions, an allowable set of roadbed engineering parameters, high-level strategies, and low-level operators, wherein the combination of roadbed parameters includes wall height, wall top width, bottom plate thickness, wall toe overhang length, wall heel overhang length, distance between the front edge of the tenon and the toe of the wall, tenon height, and tenon width;

[0013] Determine a parameter set and an objective function to be optimized for the roadbed project based on the wall height, the wall top width, the bottom plate thickness, the wall toe overhang length, the wall heel overhang length, the distance between the front edge of the tenon and the wall toe, the tenon height and the tenon width;

[0014] Establishing a roadbed engineering parameter optimization model based on the roadbed engineering parameter set to be optimized, the objective function, the geotechnical calculation function and the roadbed engineering parameter allowable set;

[0015] An adaptive hyper-heuristic algorithm is determined based on the high-level strategy and the low-level operator.

[0016] In one embodiment, the step of generating a population by performing population genetics and reverse learning based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, and determining the target population individual information comprises:

[0017] Get the greedy learning mechanism, random initial population, random factors and mutation perturbations;

[0018] The random initial population is screened by a geotechnical calculation program, the engineering cost is calculated and sorted, and the random initial population is reversely learned based on the adaptive hyper-heuristic algorithm to expand the population search space and determine the expanded population;

[0019] Screening the extended population based on the greedy learning mechanism to determine an iterative population;

[0020] Improving local search capability based on the random factor and the mutation disturbance, and determining a crossover operator and a mutation operator;

[0021] Based on the adaptive hyper-heuristic algorithm, the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the iterative population, the crossover operator and the mutation operator, population genetics is performed to select excellent individuals in the population, and repeated population genetics and reverse learning iterative training are triggered to determine the target population individual information.

[0022] In one embodiment, the step of performing population genetics to select excellent individuals based on the adaptive hyper-heuristic algorithm, the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the iterative population, the crossover operator and the mutation operator, and triggering repeated population genetics and reverse learning iterative training to determine the target population individual information includes:

[0023] Determining a current population based on the adaptive hyper-heuristic algorithm, the roadbed engineering parameter set to be optimized, and the roadbed engineering parameter optimization model;

[0024] Performing fitness evaluation based on the iterative population and the current population to determine the population cost;

[0025] Selecting excellent individuals of the population based on the population cost as the population parent set;

[0026] Performing population genetics based on the crossover operator and the population parent set to generate an excellent population, and determining an updated population;

[0027] The target population individual information is obtained based on the mutation operator and the update population.

[0028] In one embodiment, the step of adjusting the search direction and the search step length by performing correlation analysis based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm and the target population individual information to determine the target optimized population information includes:

[0029] Obtain target individual component values, target function values, target population parameter means, population standard deviations, and necessary step length sets;

[0030] Calculate the search direction based on the target population individual information and the objective function in the roadbed engineering parameter optimization model;

[0031] Calculate a Pearson correlation coefficient set based on the target individual component value, the target function value, the target population parameter mean and the population standard deviation;

[0032] Calculate the search step length based on the roadbed engineering parameter set to be optimized, the Pearson correlation coefficient set and the necessary step length set;

[0033] Adjust the search direction and the search step length based on the roadbed engineering parameter set to be optimized and the Pearson correlation coefficient set, and determine the search optimization direction and step length optimization vector;

[0034] Determining an adaptive operator based on the search optimization direction and the step size optimization vector;

[0035] Based on the adaptive operator and the target population individual information, the step gradient adaptively updates the population to obtain the target optimized population information.

[0036] In one embodiment, the step of adjusting the search direction and the search step size based on the roadbed engineering parameter set to be optimized and the Pearson correlation coefficient set, and determining the search optimization direction and step size optimization vector comprises:

[0037] Obtain the value range of roadbed engineering parameters and necessary iteration step length;

[0038] Based on the value range of the roadbed engineering parameter and the necessary iteration step, the roadbed engineering parameter set to be optimized and the search step are selected to determine the roadbed engineering optimization parameter set;

[0039] The search direction and the search step length are adjusted based on the roadbed engineering optimization parameter set and the Pearson correlation coefficient set to obtain a search optimization direction and a step length optimization vector.

[0040] In one embodiment, the step of completing intelligent optimization of roadbed structure parameters based on the target optimization population information includes:

[0041] Obtain limit iteration number and convergence conditions;

[0042] When the target optimization population information does not meet the convergence condition and does not meet the limited number of iterations, adjusting the limited number of iterations to determine the updated number of iterations, triggering repeated population training based on the updated number of iterations, and updating the target optimization population information;

[0043] When the target optimized population information satisfies the convergence condition, the target optimized population information is output.

[0044] In addition, to achieve the above purpose, the present application also proposes a roadbed structure parameter intelligent optimization device integrating population genetics and reverse learning, and the roadbed structure parameter intelligent optimization device integrating population genetics and reverse learning comprises:

[0045] An acquisition module is used to acquire a set of parameters to be optimized for a roadbed project, a roadbed project parameter optimization model, and an adaptive hyper-heuristic algorithm;

[0046] A processing module, used for generating a population by performing population genetics and reverse learning based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, and determining individual information of the target population;

[0047] The execution module is used to adjust the search direction and search step size by performing correlation analysis based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm and the target population individual information, determine the target optimization population information, and complete the intelligent optimization of the roadbed structure parameters based on the target optimization population information.

[0048] In addition, to achieve the above-mentioned objectives, the present application also proposes a device for intelligent optimization of roadbed structure parameters that integrates population genetics and reverse learning. The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. The computer program is configured to implement the steps of the method for intelligent optimization of roadbed structure parameters that integrates population genetics and reverse learning as described above.

[0049] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the intelligent optimization method of roadbed structure parameters integrating population genetics and reverse learning as described above are implemented.

[0050] One or more technical solutions proposed in this application have at least the following technical effects:

[0051] The present embodiment proposes a method for intelligent optimization of roadbed structure parameters that integrates population genetics and reverse learning, which obtains a set of roadbed engineering parameters to be optimized, a roadbed engineering parameter optimization model, and an adaptive hyper-heuristic algorithm; based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, and the adaptive hyper-heuristic algorithm, population genetics and reverse learning are performed to generate a population, and individual information of the target population is determined; based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm, and the target population individual information, correlation analysis is performed to adjust the search direction and search step length, and the target optimized population information is determined, and the intelligent optimization of roadbed structure parameters is completed based on the target optimized population information. The present application obtains a set of roadbed engineering parameters to be optimized, a roadbed engineering parameter optimization model, and an adaptive hyper-heuristic algorithm, generates a population using population genetics and reverse learning strategies and determines individual information, performs correlation analysis to adjust the search direction and step length, and repeats training to complete the intelligent optimization of roadbed structure parameters. It significantly improves the optimization efficiency and quality, and by introducing the reverse learning mechanism and dynamic reverse learning strategy, it effectively avoids local optimal solutions and obtains the global optimal solution efficiently and accurately. It has strong versatility and stability and is suitable for a variety of transportation infrastructure design fields. It significantly improves the automation and intelligence level of roadbed engineering parameter optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0053] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0054] Figure 1 A flow chart of the first embodiment of the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning provided in this application;

[0055] Figure 2 A table of correlation data for the intelligent optimization method of roadbed structure parameters integrating population genetics and reverse learning for this application;

[0056] Figure 3 A schematic diagram of cantilever retaining wall parameters and their variation ranges for the intelligent optimization method of roadbed structure parameters integrating population genetics and reverse learning for this application;

[0057] Figure 4The cantilever retaining wall size parameter diagram for the intelligent optimization method of roadbed structure parameters integrating population genetics and reverse learning for this application;

[0058] Figure 5 A flow chart of Embodiment 2 of the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning provided in this application;

[0059] Figure 6 A schematic diagram of a simplified process of an intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning provided in an embodiment of the present application;

[0060] Figure 7 Design cross-section diagram for intelligent optimization method of roadbed structure parameters integrating population genetics and reverse learning for this application;

[0061] Figure 8 The cost change diagram of the best individual in each generation of the intelligent optimization method of roadbed structure parameters that integrates population genetics and reverse learning for this application;

[0062] Fig. 9 The average cost of all individuals in each generation of the intelligent optimization method for roadbed structure parameters that integrates population genetics and reverse learning for this application;

[0063] Fig.10 Schematic diagram of different optimization results of cantilever retaining wall using intelligent optimization method for roadbed structural parameters integrating population genetics and reverse learning in this application;

[0064] Fig.11 Scatter diagram of the intelligent optimization method of roadbed structure parameters integrating population genetics and reverse learning for this application;

[0065] Fig.12 This is a convergence trend fitting curve diagram of the intelligent optimization method for roadbed structure parameters that integrates population genetics and reverse learning for this application;

[0066] Fig.13 This is a schematic diagram of the module structure of the intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning in an embodiment of the present application;

[0067] Fig.14 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning in the embodiment of the present application.

[0068] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0069] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0070] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0071] The main solution of the embodiment of the present application is: obtaining a set of parameters to be optimized for the roadbed engineering, a roadbed engineering parameter optimization model and an adaptive hyper-heuristic algorithm; generating a population by performing population genetics and reverse learning based on the set of parameters to be optimized for the roadbed engineering, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, and determining individual information of the target population; adjusting the search direction and search step by performing correlation analysis based on the set of parameters to be optimized for the roadbed engineering, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm and the individual information of the target population, determining the target optimized population information, and completing intelligent optimization of roadbed structure parameters based on the target optimized population information.

[0072] In this embodiment, for ease of description, the following description is made by taking the intelligent optimization device for roadbed structure parameters that identifies the integration of population genetics and reverse learning as the execution subject.

[0073] The manual parameter adjustment of existing technologies is time-consuming and inefficient, requiring a lot of time to try different parameter combinations, and the optimization process is slow. The selection of parameters relies on the experience and expertise of engineers, which has a high threshold. The design preferences of different engineers will cause the scheme to be affected by personal subjective factors. In addition, whether it is optimized by manual parameter adjustment or classical heuristic algorithms, it is easy to fall into the dilemma of local optimality and it is difficult to find the global optimal solution. For example, in the complex roadbed structure parameter optimization problem, the algorithm may obtain the optimal solution at a local point, but this solution is not the global optimal solution in the entire problem domain, thus limiting the optimization effect.

[0074] The present application provides a solution, which includes obtaining a set of parameters to be optimized for a roadbed engineering project, a roadbed engineering parameter optimization model, and an adaptive hyper-heuristic algorithm; performing population genetics and reverse learning based on the set of parameters to be optimized for the roadbed engineering project, the roadbed engineering parameter optimization model, and the adaptive hyper-heuristic algorithm to generate a population, and determining individual information of a target population; performing correlation analysis based on the set of parameters to be optimized for the roadbed engineering project, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm, and the individual information of the target population to adjust the search direction and the search step, determine the target optimized population information, and complete the intelligent optimization of roadbed structure parameters based on the target optimized population information.

[0075] From the above embodiments, it can be seen that the present application obtains the set of parameters to be optimized for the roadbed engineering, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, uses population genetics and reverse learning strategies to generate populations and determine individual information, performs correlation analysis to adjust the search direction and step size, and repeats training to complete the intelligent optimization of roadbed structural parameters. The optimization efficiency and quality are significantly improved, and by introducing the reverse learning mechanism and dynamic reverse learning strategy, the local optimal solution is effectively avoided, and the global optimal solution is obtained efficiently and accurately. At the same time, it has strong versatility and stability, is applicable to a variety of transportation infrastructure design fields, and significantly improves the automation and intelligence level of roadbed engineering parameter optimization.

[0076] Based on this, the embodiment of the present application provides a method for intelligent optimization of roadbed structure parameters integrating population genetics and reverse learning, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning in this application.

[0077] In this embodiment, the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning includes steps S10 to S30:

[0078] Step S10, obtaining a set of parameters to be optimized for the roadbed engineering, a roadbed engineering parameter optimization model, and an adaptive hyper-heuristic algorithm;

[0079] It should be noted that the parameter set to be optimized in the roadbed engineering reflects the characteristics of the specific parameter set that needs to be optimized in the roadbed engineering, the roadbed engineering parameter optimization model reflects the characteristics of the mathematical model established according to the safety performance requirements and economic performance requirements that need to be followed in the roadbed engineering, and the adaptive hyper-heuristic algorithm reflects the combination of multiple heuristic strategies, and automatically adapts to different optimization characteristics by dynamically adjusting algorithm parameters and search strategies.

[0080] It can be understood that the set of parameters to be optimized for the roadbed engineering is a combination of parameters whose optimal values ​​need to be determined through an optimization algorithm in the design of the roadbed engineering. It directly affects the performance, safety and economy of the roadbed structure. By optimizing the set of parameters to be optimized for the roadbed engineering, the cost can be minimized or the performance can be maximized while meeting the safety and functional requirements of the project. The roadbed engineering parameter optimization model is used to describe the relationship between various parameters in the roadbed engineering, as well as the impact of these parameters on the engineering performance and economy, converting complex engineering problems into mathematical problems, and providing a clear framework for the optimization algorithm to find the optimal solution under given constraints.

[0081] In addition, it should be noted that the adaptive hyper-heuristic algorithm may include high-level heuristic strategies and low-level heuristic operators. The high-level heuristic strategy may include a population genetic strategy and a dynamic reverse learning strategy. The population genetic strategy draws on the idea of ​​a genetic algorithm, including operations such as selection, crossover, and mutation. These operations generate a new population and gradually approach the optimal solution. The dynamic reverse learning strategy introduces a reverse learning mechanism during population evolution to generate a reverse solution, explore areas that have not yet been covered in the solution space, improve the diversity of the population, and avoid local optimality. The low-level heuristic operator may include an adaptive operator for optimizing the current population, that is, by combining and adjusting the crossover operator and the mutation operator, an efficient search of the solution space is achieved. For example, the crossover operator can generate new individuals by exchanging part of the gene information to increase the diversity of the population. The mutation operator can introduce new mutations by randomly changing certain genes of the individual to avoid the algorithm from falling into a local optimal state. Based on the fitness evaluation, excellent individuals are selected to be inherited to the next generation to ensure that the quality of the population continues to improve.

[0082] For ease of understanding, the following is explained by taking the acquisition of the parameter set to be optimized for the roadbed engineering, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm as an example, wherein the information collection device is an information collection module and the storage device is a memory.

[0083] The information collection module obtains the wall height x 1 、Wall top width x 2 、Bottom thickness x 3 、Wall toe cantilever length x 4 、Wall heel cantilever length x 5 , Distance from the front edge of the tenon to the toe of the wall x 6 , tenon height x 7 and tenon width x 8 , based on the wall height, the wall top width, the bottom plate thickness, the wall toe cantilever length, the wall heel cantilever length, the distance between the front edge of the tenon and the wall toe, the tenon height and the tenon width, determine the roadbed engineering parameter set to be optimized (x 1 ,x 2 ,…,x n ) and objective function f, and obtain the geotechnical calculation function g j (x 1 ,x 2 ,…,x n ), obtain the permissible set of roadbed engineering parameters {a i ,a i +δx i ,…,b i}(1≤i≤n), obtain high-level strategies, such as population genetic strategy and dynamic reverse learning strategy, obtain low-level operators, such as adaptive operators, and establish a roadbed engineering parameter optimization model based on the roadbed engineering parameter set to be optimized, the objective function, the geotechnical calculation function and the roadbed engineering parameter allowable set, which is abstractly described as:

[0084] y=minf(x 1 ,x 2 ,…,x n )

[0085] satisfy:

[0086]

[0087] Among them, (x 1 ,x 2 ,…,x n ) is the parameter set to be optimized for the roadbed engineering, i.e., the parameters of the retaining structure or foundation reinforcement that need to be optimized, f is the objective function, which represents the objective function that depends on the parameters, and g j (x 1 ,x 2 ,…,x n ) is the geotechnical calculation function called, which represents the safety performance requirements in the retaining structure or foundation reinforcement problem, such as the maximum bearing capacity, minimum stability requirements, etc. i ,a i +δx i ,…,b i}(1≤i≤n) represents the parameter x i The admissible set of i ,b i Respectively represent the parameters x i The minimum and maximum values ​​of δx i Represents parameter x i The minimum permissible change.

[0088] An adaptive hyper-heuristic algorithm is determined based on the high-level strategy and the low-level operator, and subsequent processing is performed based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm.

[0089] In a feasible implementation, step S10 may include steps A11 to A14:

[0090] Step A11, obtaining a combination of roadbed parameters, geotechnical calculation functions, an allowable set of roadbed engineering parameters, high-level strategies, and low-level operators, wherein the combination of roadbed parameters includes wall height, wall top width, bottom plate thickness, wall toe overhang length, wall heel overhang length, distance between the front edge of the tenon and the toe of the wall, tenon height, and tenon width;

[0091] It should be noted that the combination of roadbed parameters reflects the characteristics of the specific parameter set required in the roadbed project. The wall height refers to the vertical height of the retaining wall from the top surface of the foundation to the top of the wall, the wall top width refers to the width of the top of the retaining wall, the bottom plate thickness refers to the thickness of the bottom plate of the retaining wall, the wall toe overhang length refers to the length of the toe part of the retaining wall extending out of the foundation, the wall heel overhang length refers to the length of the heel part of the retaining wall extending out of the foundation, the distance from the front edge of the tenon to the wall toe refers to the distance from the front end of the tenon to the wall toe, the tenon height refers to the vertical height of the tenon, and the tenon width refers to the horizontal width of the tenon, the geotechnical calculation function refers to the geotechnical mechanics related function used to calculate the force and stability of the retaining wall, including settlement calculation, bearing capacity calculation, stability analysis, deformation analysis and other analyses, the allowable set of roadbed engineering parameters refers to the allowable value range of each parameter in the retaining wall design, including the minimum value, maximum value and minimum allowable change step size, the high-level strategy includes a population genetic strategy and a dynamic reverse learning strategy, which are used to guide the evolution and search direction of the population, and the low-level operator includes an adaptive operator, which is used to generate new individuals and adjust the search step size.

[0092] Step A12, determining a parameter set to be optimized and an objective function of the roadbed engineering based on the wall height, the wall top width, the bottom plate thickness, the wall toe overhang length, the wall heel overhang length, the distance between the front edge of the tenon and the wall toe, the tenon height and the tenon width;

[0093] It should be noted that the objective function reflects the characteristics of calculating and evaluating the total cost of the roadbed project under a given set of parameters.

[0094] It is understandable that the objective function needs to comprehensively consider safety indicators and economic indicators to calculate the total construction cost. The safety indicators include post-construction settlement, stability coefficient and structural deformation. The economic indicators include earthwork volume, concrete consumption and steel consumption. The introduction of the objective function can take into account both economy and safety at the same time, thereby minimizing costs and avoiding the problem of simply pursuing cost reduction while ignoring structural safety. It generates a more comprehensive and balanced design scheme, directly optimizes costs, and ensures that costs are minimized while meeting engineering safety and functional requirements, significantly improving the economic benefits of the project and enhancing the competitiveness of the design scheme in the market.

[0095] Step A13, establishing a roadbed engineering parameter optimization model based on the roadbed engineering parameter set to be optimized, the objective function, the geotechnical calculation function and the roadbed engineering parameter allowable set;

[0096] It can be understood that the established roadbed engineering parameter optimization model provides a clear framework for the optimization algorithm, so that the algorithm can find the optimal solution under given constraints, and by evaluating the output of the roadbed engineering parameter optimization model, that is, balancing cost and safety indicators, avoid the problems caused by single-objective optimization, thereby guiding the search direction and selecting the optimal parameter combination. Moreover, through automation, the time and energy of manual trial and error are significantly reduced, a large number of search options are explored in a short time, and a near-optimal solution is quickly found.

[0097] Step A14, determining an adaptive hyper-heuristic algorithm based on the high-level strategy and the low-level operator.

[0098] It can be understood that the adaptive hyper-heuristic algorithm uses population genetic strategies to generate new populations through operations such as selection, crossover, and mutation, ensuring that the algorithm continuously explores new solutions in the global search, improving the probability of finding the global optimal solution, and using a dynamic reverse learning strategy to generate reverse solutions, explore areas in the solution space that have not yet been covered, and avoid the algorithm from falling into local optimality. At the same time, the dynamic adjustment mechanism of the adaptive operator makes the algorithm more efficient in local search, significantly reducing a large amount of computing time and resource consumption.

[0099] Step S20, generating a population by performing population genetics and reverse learning based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, and determining individual information of the target population;

[0100] It should be noted that the target population individual information reflects the characteristics of the population individuals trained during the optimization process.

[0101] It can be understood that the target population individual information can characterize the individuals generated by the optimization algorithm, which is a specific combination of parameters, and each population individual has its own cost, that is, there is a fitness value. The higher the fitness value, the better the individual in the population. New individuals are generated through crossover and mutation, which increases the diversity of the population, and individuals with higher fitness are selected for reproduction, which promotes the evolution of the population and significantly improves the overall optimization effect.

[0102] For ease of understanding, the example of determining individual information of a target population is used for explanation, wherein the information collection device is an information collection module, the storage device is a memory, and the processing device is a processing module.

[0103] The information acquisition module obtains the greedy learning mechanism, random initial population, random factors and mutation disturbance, screens the random initial population through the geotechnical calculation program, calculates the engineering cost and sorts it, and performs reverse learning on the random initial population based on the adaptive hyper-heuristic algorithm to expand the population search space and determine the expanded population, that is, using the dynamic reverse learning strategy. The dynamic reverse learning strategy is aimed at the problem that the traditional genetic algorithm is limited in the global search range, easy to fall into the local optimal phenomenon and slow convergence speed due to the population diversity and insufficient local search and optimization ability in the later stage when solving the optimal parameter problem of the roadbed structure. It is proposed to integrate the dynamic reverse learning strategy into the genetic algorithm. First, the dynamic reverse learning strategy is introduced in the link of generating the initialization population. The search space is expanded and the diversity is increased through the greedy learning mechanism to improve the accuracy and convergence speed of the global optimal solution; then the random factors and mutation disturbances are added to improve the crossover and mutation operators, and improve the local search ability of the optimal solution; finally, according to the correlation analysis between the roadbed structure parameters, the adaptive operator is introduced to enhance the ability to jump out of the local optimal solution. The dynamic reverse learning strategy is used to increase the diversity of the population, and its population initialization is described as follows:

[0104] X i,j =lw j +rand·(uw j -lw j )

[0105]

[0106] Among them, X is the updated population obtained by the random normal initialization population through the GA algorithm, X DO is a population obtained by DOBL that is better than the normal iterative population solution. At any iteration t, the population size is set to N (for any individual i, i = {1, 2, ..., N}), and the solution dimension is defined as D (for any dimension j, j = {1, 2, ..., D}). [lw,uw] is the boundary of the solution, r1 i and r2 i is a random number in the range [0,1]. In addition, the population boundaries generated by DOBL need to be reviewed.

[0107] Finally, from {X∪X DO A greedy learning mechanism is introduced in} to select the first N individuals with the best fitness as the updated new initialization population X. Compared with the random initialization method, the initialization population generated by dynamic reverse learning has a better population distribution.

[0108] Based on the greedy learning mechanism, the expanded population is screened, the iterative population is determined, the local search capability is improved based on the random factor and the mutation disturbance, and the crossover operator and the mutation operator are determined, that is, the greedy learning mechanism is used. The greedy learning mechanism is based on the greedy algorithm, by screening the two most excellent elite individuals A1 and A2 in the population, and performing rotation matching learning with the remaining non-excellent individuals A3, and then using the greedy mechanism to screen out the optimal individual population within the specified range, so as to improve the global and local search capabilities, which is expressed as:

[0109] A n =A i , if(r i ·w i ∈(w 1 ,w 2 )or w i ∈(w 1 ,w 2 ))

[0110] where r i is a random number belonging to (0,1), the population size is set to M (for any individual i, i={1,2,...,M}), the total number of individuals after screening is N (n={1,2,...,N}), w 1 ,w 2 is the fitness of individuals A1 and A2.

[0111] The current population is determined based on the adaptive hyper-heuristic algorithm, the set of parameters to be optimized for the roadbed engineering and the roadbed engineering parameter optimization model; the fitness is evaluated based on the iterative population and the current population to determine the population cost; excellent individuals of the population are selected based on the population cost as the population parent set; the excellent population is generated by population genetics based on the crossover operator and the population parent set; the updated population is determined; the target population individual information is obtained based on the mutation operator and the updated population, that is, the population genetic strategy is used. The population genetic strategy mainly draws on the genetic algorithm, including encoding, iterative population generation, fitness evaluation, selection, crossover and mutation. The encoding refers to that the genetic algorithm first represents the data of the solution space as genotype string structure data of the genetic space before searching. The iterative population The generation refers to the random generation of a number of string structure data, each string structure data is called an individual, and all individuals constitute a group. The fitness evaluation refers to indicating the superiority of the individual, that is, characterizing the population cost. The selection refers to selecting excellent individuals in the current group so that they have the opportunity to reproduce as parents for the next generation. The crossover is the most important genetic operation in the genetic algorithm. A new generation of individuals can be obtained through crossover. The new individuals combine the characteristics of the ancestral individuals and embody the idea of ​​information exchange. The mutation refers to setting the mutation probability. For each individual in the group, the randomly generated probability value is compared with the set probability. The selected individual randomly changes the value of a string in the string structure data with a certain probability, thereby obtaining the target population individual information, and subsequent processing is performed based on the target population individual information.

[0112] Step S30, based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm and the target population individual information, a correlation analysis is performed to adjust the search direction and the search step, the target optimization population information is determined, and the intelligent optimization of the roadbed structure parameters is completed based on the target optimization population information.

[0113] It should be noted that the target optimized population information reflects the characteristics of the optimal population information finally outputted after continuous optimization.

[0114] It can be understood that the target optimization population information can characterize the optimal parameter combination obtained after multiple rounds of iteration and optimization of the population individuals. At this time, the best balance is achieved in economy and safety, and the cost is minimized or the performance is maximized while meeting all safety performance requirements. At the same time, the convergence state of the target optimization population information is manually determined to evaluate the efficiency and stability of the optimization process, ensuring that the algorithm finds the optimal solution within a reasonable time, making more efficient use of computing resources, reducing unnecessary calculations, and improving optimization efficiency.

[0115] For ease of understanding, the determination of target optimization population information is taken as an example for explanation, wherein the information collection device is an information collection module, the storage device is a memory, and the execution device is an execution module.

[0116] The information acquisition module obtains the target individual component value x i k , get the objective function value y k , get the target population parameter mean μx i and μy, to obtain the population standard deviation σx i and σy. Since there are many types of structures in roadbed engineering, such as retaining structures, foundation reinforcement and slope protection structures, the constraint function of its structural safety performance is g j (x 1 ,x 2 ,…,x n ) and the objective function is f(x 1 ,x 2 ,…,x n ), based on the target population individual information and the objective function in the roadbed engineering parameter optimization model, the search direction is calculated, based on the roadbed engineering parameter value range and the necessary iteration step, the roadbed engineering parameter set to be optimized and the search step are screened, the roadbed engineering optimization parameter set is determined, based on the roadbed engineering optimization parameter set and the Pearson correlation coefficient set, the search direction and the search step are adjusted to obtain the search optimization direction and the step optimization vector, based on the search optimization direction and the step optimization vector, an adaptive operator is determined, based on the adaptive operator and the target population individual information, the step gradient is adaptively updated to obtain the target optimization population information, that is, the super heuristic algorithm is used to optimize the design of the retaining structure and foundation reinforcement parameters of the foundation engineering, and the parameters containing n parameters are optimized. The roadbed structure model, let the column vector X = (x 1 ,x 2 ,…,x n ) T , where f is the objective function about X. In the algorithm, a single vector X represents an individual, and the population represents the set of individuals in the specified structure. The adaptive hyper-heuristic algorithm from the k-th step population individual to the k+1-th step population individual can be expressed as:

[0117]

[0118] in Represents an adaptive operator. The construction of the adaptive operator is determined by the adaptive search direction and the adaptive step size based on "step size + gradient", which depends on the gradient g of the objective function f. f , the step vector of the independent variable δx=(δx 1 ,δx 2 ,…,δxn ) T And other parameters p a , such as sensitivity analysis, number of parameters and problem size.

[0119] Considering adaptive operators Defined as Current population individual X k To the next generation, individual X k+1 The evolution of is expressed as:

[0120]

[0121] in Representation and The nearest allowable value. A represents the operator related to the adaptive search direction, S = (S 1 ,S 2 ,…,S n ) T It represents the step vector along the search direction. Generally, S = δx. Smin and Smax represent the minimum step and maximum step, respectively.

[0122] The search direction needs to be adaptively adjusted according to the influence of each parameter. The gradient value can be used to reflect the trend of individual fitness changes. Therefore, the search direction operator A can use the gradient of the objective function f To indicate that,

[0123] In actual calculations, A is a diagonal matrix of order n, with diagonal element a ii By the partial derivative of the objective function f Calculated. It is approximately represented by the difference quotient Δf / Δx of the previous two generations of data. For example, in the pile-sheet wall, there are three parameters: pile length, rectangular pile width, and rectangular pile spacing. During the operation of the algorithm, two optimal individuals X are obtained successively. 1 ,X 2 and the corresponding cost f 1 ,f 2 ,have

[0124] ΔX=X 2 -X 1 =(Δx 1 ,Δx 2 ,Δx 3 ) T ,

[0125] Δf=f 2 -f 1 ,

[0126] Then we can calculate

[0127]

[0128] If Δx i Not all 0, such as Δx 1 =0, then take Δf / Δx 1 =0, and Approximate value L 1 After regularization, it is used as the diagonal element of A to obtain the search direction, which is expressed as:

[0129]

[0130] If Δx i If all are 0, then take Δf / Δx i =1, then A is the unit matrix. To avoid Δx i If it is always 0, the next generation of calculations will directly perform selection, crossover, and mutation operations.

[0131] For example, the sample individual X of pile-sheet wall of roadbed retaining structure 1 =(11,1.75,5) T (Indicates that the pile length is 11m, the pile width is 1.75m, and the pile spacing is 5m), X 2 =(14,1.75,7) T (Indicates that the pile length is 14m, the rectangular pile width is 1.75m, and the pile spacing is 7m), the corresponding cost is f 1 =4860.88 (yuan / meter), f 2 =4404.51 (yuan / meter), then we can calculate And regularize the approximation to get the diagonal element of A, that is (single calculation direction)

[0132]

[0133] The Pearson correlation coefficient set is calculated based on the target individual component value, the target function value, the target population parameter mean and the population standard deviation. In order to improve the efficiency of the algorithm, the correlation analysis of parameters is introduced into each component of the commonly used step vector S, that is, the strong and weak correlation factors between variables and costs are introduced in the calculation of the step vector.

[0134] The commonly used Pearson correlation coefficient is used to indicate the degree of correlation. i The corresponding correlation coefficient is r i , then it is expressed as:

[0135]

[0136] where x i k represents the optimal individual component x of the kth generation iThe value of y k represents the optimal value of the objective function of the kth generation, μx i and μy are the optimal parameters x for all populations. i and the mean of y, σx i and σy are the optimal parameters of the population x i and the standard deviation of y and .

[0137] The search step length is calculated based on the roadbed engineering parameter set to be optimized, the Pearson correlation coefficient set and the necessary step length set, that is, the calculated correlation coefficient is r i , r i The size of reflects the strength of the correlation, such as Figure 2 As shown, Figure 2 This is the correlation data table of the intelligent optimization method of roadbed structure parameters integrating population genetics and reverse learning in this application. Intuitively, the strength of the correlation reflects the sensitivity of the objective function to the parameters, which in turn affects the design of the parameter step size, that is, using the correlation vector R = (r 1 ,r 2 ,…,r n ) T , the step vector S = (S 1 ,S 2 ,...,S n ) T It is expressed as:

[0138]

[0139] in is the minimum step size to be set; λ is a constant, which can be set to 2;

[0140] Obtain the value range of roadbed engineering parameters and the necessary iteration step, that is, preset the value range and minimum iteration step of different structural parameters before parameter optimization. Taking the cantilever retaining wall in the retaining structure as an example, the parameters and variation range are listed, such as Figure 3 As shown, Figure 3 The schematic diagram of the parameters and their variation range of cantilever retaining wall is a method for intelligent optimization of roadbed structure parameters by integrating population genetics and reverse learning for this application. Cantilever retaining walls are commonly used in station roadbeds and restricted roadbed sections. The main parameters, parameter values ​​and specific structure diagrams of the parameters are shown below. In cantilever retaining wall engineering examples, the wall height is often a known parameter controlled by other factors. Therefore, it is only necessary to construct an optimization model based on seven parameters: wall top width, bottom plate thickness, wall toe cantilever length, wall heel cantilever length, distance between the front edge of the tenon and the wall toe, tenon height and tenon width, such as Figure 4 As shown, Figure 4For the present application, a cantilever retaining wall dimension parameter diagram of an intelligent optimization method for roadbed structure parameters that integrates population genetics and reverse learning is provided, and a limit on the number of iterations and a convergence condition are obtained. When the target optimization population information does not meet the convergence condition and does not meet the limit on the number of iterations, the limit on the number of iterations is adjusted to determine the update number of iterations, and repeated population training is triggered based on the update number of iterations, and the target optimization population information is updated. When the target optimization population information meets the convergence condition, the target optimization population information is output, and the intelligent optimization of roadbed structure parameters is completed based on the target optimization population information.

[0141] In a feasible implementation, step S30 may include steps B11 to B17:

[0142] Step B11, obtaining target individual component values, target function values, target population parameter means, population standard deviations, and necessary step length sets;

[0143] It should be noted that the target individual component value reflects the characteristics of the specific component values ​​of each individual parameter in the optimization process, the objective function value reflects the characteristics of the individual fitness value calculated according to the objective function, the target population parameter mean reflects the characteristics of the average value of a certain parameter of all individuals in the current population, the population standard deviation reflects the characteristics of the discrete degree of a certain parameter value of all individuals in the current population, and the necessary step set reflects the characteristics of the minimum step set in the optimization process.

[0144] It can be understood that the target individual component value can be specific structural parameter values ​​such as wall top width, bottom plate thickness, wall toe cantilever length, such as wall top width of 0.7 meters and bottom plate thickness of 0.6 meters. It is the characteristic of each individual in the optimization algorithm and is used to evaluate the fitness of the individual. The objective function value can be the objective function, that is, the calculated engineering cost under a given parameter combination, which is used to evaluate the economy of the individual under the current parameter combination. The target population parameter mean is the average value of the wall top width of all individuals in the current population, which is used to evaluate the average level of the population as a whole on this parameter, so as to understand the distribution of the population. The population standard deviation can characterize the distribution of the wall top width parameter in the current population, which is used to evaluate the diversity of the population and avoid premature convergence to the local optimal solution during the search process.

[0145] Step B12, calculating a search direction based on the target population individual information and the objective function in the roadbed engineering parameter optimization model;

[0146] It should be noted that the search direction reflects the characteristics of the direction of the next search determined during the optimization process.

[0147] It can be understood that in the optimization of roadbed structure parameters, the search direction can represent the direction of parameter adjustment, that is, adjusting the parameter value to improve the fitness of the individual, such as reducing the cost or improving the performance.

[0148] Step B13, calculating a Pearson correlation coefficient set based on the target individual component value, the target function value, the target population parameter mean and the population standard deviation;

[0149] It should be noted that the Pearson correlation coefficient set reflects the characteristics of the set of Pearson correlation coefficients between each parameter and the objective function value calculated during the optimization process.

[0150] It can be understood that the Pearson correlation coefficient set can be used to measure the degree of linear correlation between two variables. Its value range is between -1 and 1, which characterizes the influence of each parameter on the objective function value. If the Pearson correlation coefficient of a parameter is close to 1, it means that the parameter is positively correlated with the objective function value, that is, when the parameter value increases, the objective function value also increases. If the Pearson correlation coefficient is close to -1, it means that the parameter is negatively correlated with the objective function value, that is, when the parameter value increases, the objective function value decreases.

[0151] Step B14, calculating the search step length based on the roadbed engineering parameter set to be optimized, the Pearson correlation coefficient set and the necessary step length set;

[0152] It should be noted that the search step size reflects the feature of dynamically adjusting the search step size of each parameter by using the Pearson correlation coefficient set during the optimization process.

[0153] It can be understood that the search step size determines the amplitude of parameter adjustment in each iteration. The step size is reasonably adjusted by evaluating the influence of each parameter on the objective function. If a parameter is highly positively correlated with the objective function, a smaller step size may be more appropriate during the optimization process to avoid excessive adjustment leading to deterioration of the objective function value. Conversely, if the correlation is weak, a larger step size can be used to speed up the search.

[0154] Step B15, adjusting the search direction and the search step size based on the roadbed engineering parameter set to be optimized and the Pearson correlation coefficient set, and determining the search optimization direction and step size optimization vector;

[0155] It should be noted that the search optimization direction reflects the characteristics of the final search direction adjusted during the optimization process, and the step size optimization vector reflects the characteristics of the vector composed of the search step sizes of each parameter adjusted during the optimization process.

[0156] In a feasible implementation, step B15 may include steps C11 to C13:

[0157] Step C11, obtaining the value range of roadbed engineering parameters and the necessary iteration step length;

[0158] It should be noted that the value range of the roadbed engineering parameter reflects the characteristics of the minimum and maximum values ​​allowed for each parameter in the roadbed structure design, and the necessary iteration step reflects the characteristics of the minimum adjustment range allowed for each parameter in each iteration during the optimization process.

[0159] Step C12, screening the roadbed engineering parameter set to be optimized and the search step length based on the roadbed engineering parameter value range and the necessary iteration step length, and determining the roadbed engineering optimization parameter set;

[0160] It should be noted that the roadbed engineering optimization parameter set reflects the characteristics of the parameter set obtained after screening according to the restriction conditions.

[0161] It is understandable that a reasonable setting of the parameter value range and iteration step size can maintain stability during the search process, avoid optimization failures caused by parameter values ​​exceeding a reasonable range or improper step size settings, and while ensuring optimization accuracy, accelerate convergence, reduce unnecessary calculations, improve optimization efficiency, better adapt to various optimization problems, and improve versatility.

[0162] Step C13, adjusting the search direction and the search step size based on the roadbed engineering optimization parameter set and the Pearson correlation coefficient set to obtain a search optimization direction and step size optimization vector.

[0163] It can be understood that the search optimization direction can point to the direction that is most likely to reduce the objective function value. For example, if a parameter is negatively correlated with the objective function, then the optimization direction will point to the direction in which the parameter value increases. If it is positively correlated, then it will point to the direction in which the parameter value decreases, ensuring that each step moves towards the optimization goal.

[0164] In addition, it should be noted that each element in the step size optimization vector represents the amplitude by which the corresponding parameter should be adjusted in this iteration. The importance of the parameter is measured by the Pearson correlation coefficient, and the current value range of each parameter is limited to ensure that the adjustment of each parameter is neither too radical nor too conservative. For parameters with greater influence, a smaller step size is used for fine adjustment, and for parameters with less influence, a larger step size is used to speed up the search.

[0165] Step B16, determining an adaptive operator based on the search optimization direction and the step size optimization vector;

[0166] It should be noted that the adaptive operator reflects the characteristics of adaptively adjusting the search strategy according to the operator dynamically adjusted according to the search optimization direction and the step size optimization vector.

[0167] It can be understood that the adaptive operator is used to update the parameter values ​​of the individuals in the population in each iteration to more efficiently approach the optimal solution.

[0168] Step B17, performing step gradient adaptive updating of the population based on the adaptive operator and the target population individual information to obtain target optimized population information.

[0169] It can be understood that the target optimization population information can characterize the optimal parameter combination obtained after multiple rounds of iteration and optimization of the population individuals, thereby minimizing the cost or maximizing the performance. The convergence state of the target optimization population information is manually determined to evaluate the efficiency and stability of the optimization process, thereby ensuring that the algorithm finds the optimal solution within a reasonable time, making more efficient use of computing resources, reducing unnecessary calculations, and improving optimization efficiency.

[0170] In another feasible implementation, step S30 may include steps D11 to D13:

[0171] Step D11, obtaining the limit iteration number and convergence condition;

[0172] It should be noted that the limit on the number of iterations reflects the characteristics of the initial number of iterations of the algorithm that is preset when the optimization algorithm starts to be executed, and the convergence condition reflects the characteristics of the conditions that are manually set during the optimization process to meet the optimal solution or close to the optimal solution.

[0173] It is understandable that both the limiting number of iterations and the convergence condition need to be set manually, and both may be set or only one may be set.

[0174] In addition, it should be noted that the limit on the number of iterations is used to control the algorithm to iterate at the beginning, so as to quickly explore the solution space in the early stage and find an area that is closer to the optimal solution. The convergence condition is used to determine whether the algorithm has found the optimal solution or is close to the optimal solution. When the algorithm meets the convergence condition, it is considered to have converged and the iteration can be stopped.

[0175] Step D12, when the target optimization population information does not meet the convergence condition and does not meet the limited number of iterations, adjusting the limited number of iterations to determine the updated number of iterations, triggering repeated population training based on the updated number of iterations, and updating the target optimization population information;

[0176] It should be noted that the update iteration number reflects the characteristics of a new iteration number determined after adjusting the limit iteration number when the current target optimization population information does not meet the convergence condition.

[0177] It can be understood that the update iteration number is used to control the algorithm to continue iterating based on the current iteration so as to further optimize the objective function value.

[0178] Step D13: when the target optimized population information satisfies the convergence condition, output the target optimized population information.

[0179] It can be understood that by outputting the target optimization population information when the target optimization population information meets the convergence conditions, the reliability and effectiveness of the optimization results can be ensured, and the optimization can be stopped in time when the optimization target is reached, thereby avoiding unnecessary calculations, improving the optimization efficiency, and running stably, avoiding optimization failures caused by insufficient or excessive iterations.

[0180] The present embodiment proposes a method for intelligent optimization of roadbed structure parameters integrating population genetics and reverse learning, which obtains a set of roadbed engineering parameters to be optimized, a roadbed engineering parameter optimization model and an adaptive hyper-heuristic algorithm; based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, population genetics and reverse learning are performed to generate a population, and target population individual information is determined; based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm and the target population individual information, correlation analysis is performed to adjust the search direction and search step size, determine the target optimization population information, and complete the intelligent optimization of roadbed structure parameters based on the target optimization population information. The technical problem of how to more efficiently perform intelligent optimization of roadbed structure parameters is solved. Compared with the existing technology, the present application performs population genetics and reverse learning, generates a population and determines the individual information of the target population, and adjusts the search direction and step size through correlation analysis to determine the target optimized population information, thereby completing the intelligent optimization of roadbed structure parameters, significantly improving the optimization efficiency, and enhancing the robustness and stability of the algorithm, avoiding local optimal solutions, and ensuring the reliability and effectiveness of the global optimal solution.

[0181] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction and will not be repeated later.

[0182] In this embodiment, refer to Figure 5 , Figure 5 The flowchart of the second embodiment of the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning is provided in this application, and step S20 specifically includes steps S21 to S25:

[0183] Step S21, obtaining a greedy learning mechanism, a random initial population, a random factor, and a mutation disturbance;

[0184] It should be noted that the greedy learning mechanism reflects the characteristics of the best individuals in the screening population, the random initial population reflects the characteristics of a set of initial solutions or parameter combinations randomly generated at the beginning of the optimization algorithm, the random factor reflects the random elements introduced in the optimization process to increase the exploration ability of the algorithm, and the variation disturbance reflects the random disturbance of individual parameter values ​​in the optimization process to introduce new variations.

[0185] For ease of understanding, the greedy learning mechanism, random initial population, random factors and mutation disturbance are taken as an example for explanation, wherein the information collection device is the information collection module, the storage device is the memory, and the processing device is the processing module.

[0186] The information acquisition module obtains the greedy learning mechanism, the random initial population, the random factor and the mutation disturbance. The greedy learning mechanism is used to select the current optimal solution or parameter combination in the optimization process. By screening the best individuals in the population, and performing rotation matching learning with the remaining non-excellent individuals, the greedy mechanism is used to screen out the optimal individual population within the specified range. For example, the two individuals with the highest fitness in the population, namely the elite individuals, can be selected, and crossover and mutation operations can be performed with the remaining individuals to generate a new population. Then, the individuals with the highest fitness can be screened out again to form a new population, which can quickly improve the overall fitness of the population and accelerate the convergence speed of the algorithm. The random initial population is a set of initial solutions or parameter combinations randomly generated at the beginning of the optimization algorithm, and is randomly generated within the range of parameter values. For example, for parameters such as the top width of the wall and the thickness of the bottom plate, the parameter values ​​of each individual in the random initial population are randomly selected within the preset range of values, so as to widely explore the solution space and avoid falling into the local optimal solution. The random factor can be a random number or random probability, which is used to control certain operations in the algorithm, such as crossover and mutation. For example, in the crossover operation, the random factor can determine which individuals to cross and the location of the crossover point; in the mutation operation, the random factor can determine which individuals to mutate and the magnitude of the mutation, significantly increasing the diversity and robustness of the algorithm and preventing the algorithm from converging to the local optimal solution too early. The mutation perturbation is used to introduce new mutations, increase the diversity of the population, and generate new individuals by randomly adjusting certain parameter values ​​of individuals. For example, a parameter of an individual can be randomly selected and its value increased or decreased by a random small amplitude value. The mutation perturbation helps the algorithm to jump out of the local optimal solution and explore the uncovered areas in the solution space, significantly improving the global search ability of the algorithm, and performing subsequent processing based on the greedy learning mechanism, random initial population, random factors and mutation perturbation.

[0187] Step S22, screening the random initial population through a geotechnical calculation program, calculating the engineering cost for sorting, and performing reverse learning on the random initial population based on the adaptive hyper-heuristic algorithm to expand the population search space and determine the expanded population;

[0188] It should be noted that the extended population reflects the characteristics of a new population individual set generated through a reverse learning mechanism on the basis of the original random initial population.

[0189] It can be understood that the expanded population can cover a wider solution space, increase the diversity and exploration ability of the population, and thus search for the optimal solution more comprehensively.

[0190] For ease of understanding, an explanation is given by taking determining an extended population as an example, wherein the information collection device is an information collection module, the storage device is a memory, and the processing device is a processing module.

[0191] The information acquisition module obtains the greedy learning mechanism, the random initial population, the random factors and the variation disturbance, screens the random initial population through the geotechnical calculation program, calculates the engineering cost for sorting, and performs reverse learning on the random initial population based on the adaptive hyper-heuristic algorithm to expand the population search space and determine the expanded population, that is, adopts a dynamic reverse learning strategy to increase the population diversity. The population initialization is described as follows:

[0192] X i,j =lw j +rand·(uw j -lw j )

[0193]

[0194] Among them, X is the updated population obtained by the random normal initialization population through the GA algorithm, X DO is a population obtained by DOBL that is better than the normal iterative population solution. At any iteration t, the population size is set to N (for any individual i, i = {1, 2, ..., N}), and the solution dimension is defined as D (for any dimension j, j = {1, 2, ..., D}). [lw,uw] is the boundary of the solution, r1 i and r2 i is a random number in [0,1]. In addition, the population boundary generated by DOBL needs to be reviewed to obtain an extended population, and subsequent processing is performed based on the extended population.

[0195] Step S23, screening the extended population based on the greedy learning mechanism to determine the iterative population;

[0196] It should be noted that the iterative population reflects the characteristics of the set of individuals with the best fitness selected in the extended population through the greedy learning mechanism.

[0197] For ease of understanding, an explanation is given by taking determining an iterative population as an example, wherein the information collection device is an information collection module, the storage device is a memory, and the processing device is a processing module.

[0198] The information acquisition module obtains the greedy learning mechanism, the random initial population, the random factors and the variation disturbance, selects the random initial population through the geotechnical calculation program, calculates the engineering cost for sorting, and performs reverse learning on the random initial population based on the adaptive hyper-heuristic algorithm to expand the population search space, determines the extended population, selects the extended population based on the greedy learning mechanism, determines the iterative population, that is, from {X∪X DO} introduces a greedy learning mechanism to select the first N individuals with the best fitness as the new initialization population X after updating. Compared with the random initialization method, the initialization population generated by dynamic reverse learning has a better population distribution. The greedy learning mechanism is based on the greedy algorithm. By selecting the two best elite individuals A1 and A2 in the population, they are matched with the remaining non-excellent individuals A3 for round-robin learning, and then the greedy mechanism is used to select the best individual population within the specified range, so as to improve the global and local search capabilities, which is expressed as:

[0199] A n =A i , if(r i ·w i ∈(w 1 ,w 2 )or w i ∈(w 1 ,w 2 ))

[0200] where r i is a random number belonging to (0,1), the population size is set to M (for any individual i, i={1,2,...,M}), the total number of individuals after screening is N (n={1,2,...,N}), w 1 ,w 2 is the fitness of individuals A1 and A2.

[0201] Step S24, improving the local search capability based on the random factor and the mutation disturbance, and determining a crossover operator and a mutation operator;

[0202] It should be noted that the crossover operator reflects the characteristics of generating new offspring individuals by exchanging part of the gene information of two parent individuals, and the mutation operator reflects the characteristics of introducing new mutations by randomly changing certain gene values ​​of individuals.

[0203] It can be understood that the crossover operator and mutation operator are an operation in the genetic algorithm, which is used to increase the diversity of the population, promote the exchange and combination of information, thereby generating a better solution, and increase the diversity of the population, avoid the algorithm from falling into a local optimal solution, and increase the exploration ability of the algorithm.

[0204] For ease of understanding, the determination of a crossover operator and a mutation operator is taken as an example for explanation, wherein the information collection device is an information collection module, the storage device is a memory, and the processing device is a processing module.

[0205] The information acquisition module obtains random factors and mutation disturbances, improves crossover operators and mutation operators based on the random factors and mutation disturbances, and performs subsequent processing based on the crossover operators and mutation operators.

[0206] Step S25, performing population genetics to select excellent individuals based on the adaptive hyper-heuristic algorithm, the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the iterative population, the crossover operator and the mutation operator, and triggering repeated population genetics and reverse learning iterative training to determine the target population individual information.

[0207] It should be noted that the target population individual information reflects the characteristics of the population individuals generated by repeated population genetics and reverse learning iterative training.

[0208] For ease of understanding, the example of determining individual information of a target population is used for explanation, wherein the information collection device is an information collection module, the storage device is a memory, and the processing device is a processing module.

[0209] The information acquisition module obtains an adaptive hyper-heuristic algorithm, a set of parameters to be optimized for the roadbed engineering, a model for optimizing the parameters of the roadbed engineering, an iterative population, a crossover operator and a mutation operator, determines the current population based on the adaptive hyper-heuristic algorithm, the set of parameters to be optimized for the roadbed engineering and the model for optimizing the parameters of the roadbed engineering, performs fitness evaluation based on the iterative population and the current population, determines the population cost, selects excellent individuals of the population based on the population cost as the population parent set, performs population genetics based on the crossover operator and the population parent set to generate an excellent population, determines an updated population, and obtains individual information of the target population based on the mutation operator and the updated population, i.e., utilizes a population genetic strategy, which mainly draws on a genetic algorithm, including encoding, generation of iterative populations, fitness evaluation, selection, crossover and mutation, wherein the encoding refers to that the genetic algorithm first solves the empty population before searching. The data between the two is expressed as genotype string structure data of the genetic space, the generation of the iterative population refers to the random generation of several string structure data, each string structure data is called an individual, and all individuals constitute a group, the fitness evaluation refers to indicating the superiority of the individual, that is, characterizing the population cost, the selection refers to selecting excellent individuals in the current group so that they have the opportunity to reproduce as parents for the next generation, the crossover is the most important genetic operation in the genetic algorithm, and a new generation of individuals can be obtained through crossover. The new individuals combine the characteristics of the ancestral individuals and embody the idea of ​​information exchange, and the mutation refers to setting the mutation probability. For each individual in the group, the randomly generated probability value is compared with the set probability, and the selected individual randomly changes the value of a string in the string structure data with a certain probability, thereby obtaining the target population individual information, and subsequent processing is performed based on the target population individual information.

[0210] In a feasible implementation, step S25 may include steps E11 to E15:

[0211] Step E11, determining a current population based on the adaptive hyper-heuristic algorithm, the roadbed engineering parameter set to be optimized, and the roadbed engineering parameter optimization model;

[0212] It should be noted that the current population reflects the characteristics of the optimization state of the current iteration stage after multiple iterations during the optimization process.

[0213] It can be understood that the current population can characterize the optimization state of the current iteration stage, generate high-quality individuals more quickly, improve optimization efficiency, and converge to the optimal solution more quickly.

[0214] Step E12, performing fitness evaluation based on the iterative population and the current population to determine the population cost;

[0215] It should be noted that the population cost reflects the characteristics of the cost of all individuals in the current population calculated through fitness evaluation during the optimization process.

[0216] It can be understood that the population cost is used to help the algorithm select more economical individuals for genetic operations, thereby minimizing costs while meeting safety performance requirements.

[0217] Step E13, selecting excellent individuals in the population based on the population cost as the population parent set;

[0218] It should be noted that the population parent set reflects the characteristics of a set of individuals with higher fitness selected through fitness evaluation during the optimization process.

[0219] It can be understood that the population parent set can be used to participate in crossover and mutation operations to generate new offspring individuals, and the quality of the population parent set directly affects the optimization effect of the next generation population.

[0220] Step E14, performing population genetics based on the crossover operator and the population parent set to generate an excellent population, and determining an updated population;

[0221] It should be noted that the updated population reflects the characteristics of a new population individual set generated by performing a crossover operation on individuals in the population parent set using a crossover operator.

[0222] It can be understood that the updated population inherits the excellent genes of the parent individuals, and at the same time produces new gene combinations through gene recombination, which increases the diversity of the population, thereby exploring a wider solution space.

[0223] Step E15, obtaining target population individual information based on the mutation operator and the update population.

[0224] It is understandable that the mutation operator can introduce new mutations by randomly changing the gene values ​​of individuals, further increasing the population diversity of the updated population, ensuring that the algorithm can explore a wider solution space, avoiding falling into the local optimal solution, and to a certain extent breaking the uniformity of the population, increasing the algorithm's exploration ability, enabling the algorithm to better adapt to complex optimization problems, improving the algorithm's global search ability, better adapting to different working conditions and constraints, enhancing the algorithm's robustness and stability, and ensuring that the algorithm can stably search for the global optimal solution.

[0225] The present embodiment proposes an intelligent optimization method for roadbed structure parameters that integrates population genetics and reverse learning, which obtains a greedy learning mechanism, a random initial population, a random factor, and a mutation disturbance; the random initial population is screened through a geotechnical calculation program, the engineering cost is calculated and sorted, and the random initial population is reversely learned based on the adaptive hyper-heuristic algorithm to expand the population search space and determine the extended population; the extended population is screened based on the greedy learning mechanism to determine the iterative population; the local search capability is improved based on the random factor and the mutation disturbance, and a crossover operator and a mutation operator are determined; based on the adaptive hyper-heuristic algorithm, the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the iterative population, the crossover operator, and the mutation operator, population genetics are performed to select excellent individuals of the population, and repeated population genetics and reverse learning iterative training are triggered to determine the individual information of the target population. The technical problem of how to more accurately perform intelligent optimization of roadbed structure parameters is solved. Compared with the existing technology, the present application obtains a greedy learning mechanism, a random initial population, random factors and mutation disturbances, thereby performing reverse learning on the random initial population, expanding the population search space, determining the extended population, screening the extended population through the greedy learning mechanism, determining the iterative population, and utilizing random factors and mutation disturbances to enhance the local search capability, determine the crossover operator and mutation operator, perform population inheritance, select excellent individuals, determine the target population individual information, enable the algorithm to accurately locate the optimal solution, generate higher quality population individuals, achieve cost minimization or performance maximization, and significantly improve the accuracy and reliability of population training.

[0226] For example, in order to help understand the implementation process of the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning obtained by combining this embodiment with the above-mentioned embodiment 1, please refer to Figure 6 , Figure 6 A brief flow chart of an intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning is provided. Specifically:

[0227] Corresponding to the classical genetic algorithm, in the algorithm design, this application not only retains the characteristics of the genetic algorithm in the process of generating a new generation of population, but also introduces population screening based on reverse learning and individual updating based on data analysis, that is: every several generations, a part of the new individuals are obtained by selection, crossover, and mutation, and a part of the individuals are updated by a search algorithm based on data feature analysis, and the updated population is further screened by a dynamic reverse learning strategy. In the data feature analysis, data-based analysis and calculation of difference quotients and correlation analysis are used to determine the search direction and search step.

[0228] That is, set the relevant parameters and constraints of the problem, randomly generate m parameters and s individuals m×sAs an iterative population, fitness evaluation is performed. The higher the fitness, the greater the chance of reproduction. The fitness value F=1 / f is defined, which is the inverse of the objective function f. For newly generated individuals, the external "geotechnical calculation program" is called to calculate its fitness. Based on fitness evaluation, excellent individuals in the population are selected with a certain probability to be inherited to the next generation of population. The roulette method and elite selection strategy are adopted to ensure that the excellent individuals of the previous generation are not destroyed. The gradient operator is used to guide the search direction, and the step vector determines the length of advancement. When the evolutionary generation i is an integer multiple of a given positive integer modulus n, an adaptive operation is performed to calculate the difference quotient of the first several excellent individuals in the current population and perform correlation analysis. According to the calculation and analysis results, the search direction A and the search step S are adaptively adjusted. Perform crossover and mutation of the population. According to the crossover probability p c and mutation probability p m Perform single-point crossover and uniform mutation operations on the population. Use the newly generated individuals to update the solution space. Perform dynamic reverse learning population initialization. When the calculation does not reach convergence, apply the DOBL operator to the updated population again to generate a better population, and perform population boundary review. Then, all individuals in the population that meet the conditions are screened through the greedy learning screening mechanism, and a fixed number of individuals with the best fitness are selected to form a new population. Determine the termination condition of the algorithm. If the convergence condition is met and the number of iterations does not exceed the maximum number of iterations, the optimal output of each generation is output and the algorithm ends. Otherwise, continue training iterations.

[0229] Referring to Example 1, a set of parameters to be optimized for the roadbed engineering, a roadbed engineering parameter optimization model, and an adaptive hyper-heuristic algorithm are obtained; based on the set of parameters to be optimized for the roadbed engineering, the roadbed engineering parameter optimization model, and the adaptive hyper-heuristic algorithm, population genetics and reverse learning are performed to generate a population, and individual information of the target population is determined; based on the set of parameters to be optimized for the roadbed engineering, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm, and the individual information of the target population, correlation analysis is performed to adjust the search direction and search step size, determine the target optimized population information, and complete the intelligent optimization of the roadbed structure parameters based on the target optimized population information. Referring to Example 2, a greedy learning mechanism, a random initial population, a random factor and a mutation disturbance are obtained; the random initial population is screened through a geotechnical calculation program, the engineering cost is calculated and sorted, and the random initial population is reversely learned based on the adaptive hyper-heuristic algorithm to expand the population search space and determine the extended population; the extended population is screened based on the greedy learning mechanism to determine the iterative population; the local search capability is improved based on the random factor and the mutation disturbance, and a crossover operator and a mutation operator are determined; based on the adaptive hyper-heuristic algorithm, the set of parameters to be optimized for the roadbed project, the roadbed project parameter optimization model, the iterative population, the crossover operator and the mutation operator, excellent individuals of the population are selected through population genetics, and repeated population genetics and reverse learning iterative training are triggered to determine the individual information of the target population.

[0230] like Figure 7 As shown, Figure 7 The cross-section diagram is designed for the intelligent optimization method of roadbed structure parameters integrating population genetics and reverse learning for this application. The design speed is 200km / h double-track ballasted Class I railway. The design cross-section strata are mainly deep Quaternary silty clay, silty silty clay, silty sand, etc. The left side of the cross-section adopts C35 reinforced concrete cantilever retaining wall with a wall height of 5.8m. The backfill behind the wall adopts A, B, and C group fillers. The comprehensive internal friction angle The friction coefficient of the base is f=0.5, and the top of the wall considers the Class I railway standard train load. The artificial design uses the limit state method to calculate the cantilever wall's anti-sliding stability (safety factor ≥1.30), anti-overturning stability (safety factor ≥1.60), wall section bearing capacity and other safety indicators, and conducts reinforcement design. The wall top width is selected as 0.7m, the bottom plate thickness is 0.6m, the wall toe cantilever length is 0.5m, the wall heel cantilever length is 3.6m, the mortise height is 0m, and the mortise width is 0m. Economic indicators: concrete unit price is 450 yuan / m3, and steel unit price is 4800 yuan / t. Figure 7 It can be seen that the design section wall height is 5.8m, that is, x1 = 5.8. In engineering examples, the wall height is often a known parameter controlled by the fill height. Therefore, it is only necessary to construct an optimization model based on seven parameters: wall top width, bottom plate thickness, wall toe overhang length, wall heel overhang length, distance from the front edge of the tenon to the wall toe, tenon height, and tenon width. Figure 8 As shown, Figure 8 The cost change diagram of the best individual in each generation of the subgrade structure parameter intelligent optimization method integrating population genetics and reverse learning is used for calculation optimization by using a classical genetic algorithm, integrating population genetics and reverse learning subgrade structure parameter intelligent optimization method. The adaptive operator is updated every 5 generations to obtain the cost change of the best individual in each generation of the subgrade structure, such as Fig. 9 As shown, Fig. 9 For this application, a graph showing the average cost of all individuals in each generation of an intelligent optimization method for roadbed structure parameters that integrates population genetics and reverse learning is used. A classical genetic algorithm is used to perform calculation optimization using an intelligent optimization method for roadbed structure parameters that integrates population genetics and reverse learning. The adaptive operator is updated every five generations to obtain the average cost of all individuals in each generation.

[0231] like Fig.10 As shown, Fig.10 Schematic diagram of different optimization results of cantilever retaining wall for intelligent optimization method of roadbed structure parameters integrating population genetics and reverse learning in this application. Fig.10 It can be seen that the intelligent optimization algorithm that integrates population genetics and reverse learning takes 3 minutes to optimize the cantilever retaining wall, which is 1 / 8 of the manual design and 40% less than the classical genetic algorithm; the cost after optimization is reduced by 44.7% compared with the manual design and 2.3% compared with the classical genetic algorithm. It can be seen that the optimization algorithm also has a good application effect in the optimization of cantilever retaining wall parameters, and has obvious advantages in optimization quality and efficiency. In order to further verify the performance of the adaptive hyper-heuristic algorithm proposed in the present invention in terms of robustness and stability, 100 independent solution attempts were made using the classical genetic algorithm and the method of this patent. A scatter plot of the average cost of each generation of the population in each solution process is drawn, and the convergence trend of the scatter plot is fitted into a curve, as shown in the figure. Fig.11 As shown, Fig.11 The scatter plot of the intelligent optimization method of roadbed structure parameters integrating population genetics and reverse learning for this application is shown in Figure 2. Fig.12 As shown, Fig.12 This is the convergence trend fitting curve of the intelligent optimization method of roadbed structure parameters that integrates population genetics and reverse learning in this application, combined with Fig.11 and Fig.12 The scatter plots and fitting curves of the two methods for running the cantilever retaining wall 100 times can be used to intuitively compare the solution effects of the three algorithms. After comparative analysis, it can be seen that the intelligent optimization method that integrates population genetics and reverse learning proposed in this application is significantly superior to the classical genetic algorithm in convergence rate and optimization quality, and has good stability.

[0232] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the intelligent optimization method of roadbed structure parameters integrating population genetics and reverse learning in the present application. More forms of simple transformations based on this technical concept are all within the scope of protection of the present application.

[0233] This application also provides a roadbed structure parameter intelligent optimization device integrating population genetics and reverse learning, please refer to Fig.13 The intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning includes:

[0234] An acquisition module 10 is used to acquire a set of parameters to be optimized for a roadbed engineering, a roadbed engineering parameter optimization model, and an adaptive hyper-heuristic algorithm;

[0235] A processing module 20 is used to generate a population by performing population genetics and reverse learning based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, and determine the target population individual information;

[0236] The execution module 30 is used to adjust the search direction and search step size by performing correlation analysis based on the set of parameters to be optimized for the roadbed engineering, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm and the target population individual information, determine the target optimization population information, and complete the intelligent optimization of the roadbed structure parameters based on the target optimization population information.

[0237] The acquisition module 10 is further used to acquire a combination of roadbed parameters, geotechnical calculation functions, an allowable set of roadbed engineering parameters, high-level strategies and low-level operators, wherein the combination of roadbed parameters includes wall height, wall top width, bottom plate thickness, wall toe overhang length, wall heel overhang length, distance between the front edge of the tenon and the toe of the wall, tenon height and tenon width;

[0238] Determine a parameter set and an objective function to be optimized for the roadbed project based on the wall height, the wall top width, the bottom plate thickness, the wall toe overhang length, the wall heel overhang length, the distance between the front edge of the tenon and the wall toe, the tenon height and the tenon width;

[0239] Establishing a roadbed engineering parameter optimization model based on the roadbed engineering parameter set to be optimized, the objective function, the geotechnical calculation function and the roadbed engineering parameter allowable set;

[0240] An adaptive hyper-heuristic algorithm is determined based on the high-level strategy and the low-level operator.

[0241] The processing module 20 is also used to obtain a greedy learning mechanism, a random initial population, a random factor and a mutation disturbance;

[0242] The random initial population is screened by a geotechnical calculation program, the engineering cost is calculated and sorted, and the random initial population is reversely learned based on the adaptive hyper-heuristic algorithm to expand the population search space and determine the expanded population;

[0243] Screening the extended population based on the greedy learning mechanism to determine an iterative population;

[0244] Enhance local search capability based on the random factor and the mutation disturbance, and determine a crossover operator and a mutation operator;

[0245] Based on the adaptive hyper-heuristic algorithm, the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the iterative population, the crossover operator and the mutation operator, population genetics is performed to select excellent individuals in the population, and repeated population genetics and reverse learning iterative training are triggered to determine the target population individual information.

[0246] The processing module 20 is further used to determine the current population based on the adaptive hyper-heuristic algorithm, the roadbed engineering parameter set to be optimized and the roadbed engineering parameter optimization model;

[0247] Performing fitness evaluation based on the iterative population and the current population to determine the population cost;

[0248] Selecting excellent individuals of the population based on the population cost as the population parent set;

[0249] Performing population genetics based on the crossover operator and the population parent set to generate an excellent population, and determining an updated population;

[0250] The target population individual information is obtained based on the mutation operator and the update population.

[0251] The execution module 30 is also used to obtain target individual component values, target function values, target population parameter means, population standard deviations and necessary step sets;

[0252] Calculate the search direction based on the target population individual information and the objective function in the roadbed engineering parameter optimization model;

[0253] Calculate a Pearson correlation coefficient set based on the target individual component value, the target function value, the target population parameter mean and the population standard deviation;

[0254] Calculate the search step length based on the roadbed engineering parameter set to be optimized, the Pearson correlation coefficient set and the necessary step length set;

[0255] Adjust the search direction and the search step length based on the roadbed engineering parameter set to be optimized and the Pearson correlation coefficient set, and determine the search optimization direction and step length optimization vector;

[0256] Determining an adaptive operator based on the search optimization direction and the step size optimization vector;

[0257] Based on the adaptive operator and the target population individual information, the step gradient adaptively updates the population to obtain the target optimized population information.

[0258] The execution module 30 is also used to obtain the value range of the roadbed engineering parameters and the necessary iteration step length;

[0259] Based on the value range of the roadbed engineering parameter and the necessary iteration step, the roadbed engineering parameter set to be optimized and the search step are selected to determine the roadbed engineering optimization parameter set;

[0260] The search direction and the search step length are adjusted based on the roadbed engineering optimization parameter set and the Pearson correlation coefficient set to obtain a search optimization direction and a step length optimization vector.

[0261] The execution module 30 is also used to obtain the limit number of iterations and the convergence condition;

[0262] When the target optimization population information does not meet the convergence condition and does not meet the limited number of iterations, adjusting the limited number of iterations to determine the updated number of iterations, triggering repeated population training based on the updated number of iterations, and updating the target optimization population information;

[0263] When the target optimized population information satisfies the convergence condition, the target optimized population information is output.

[0264] The intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning provided by the present application adopts the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning in the above-mentioned embodiment, which can solve the technical problem of how to intelligently optimize roadbed structure parameters more efficiently and accurately. Compared with the prior art, the beneficial effects of the intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning provided by the present application are the same as the beneficial effects of the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning provided by the above-mentioned embodiment, and the other technical features of the intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning are the same as the features disclosed in the above-mentioned embodiment method, which will not be repeated here.

[0265] The present application provides an intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning. The intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning in the above-mentioned embodiment one.

[0266] Reference below Fig.14 , which shows a schematic diagram of the structure of a roadbed structure parameter intelligent optimization device that integrates population genetics and reverse learning and is suitable for implementing the embodiment of the present application. The roadbed structure parameter intelligent optimization device that integrates population genetics and reverse learning in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions: tablet computers), PMPs (Portable Media Players: portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Fig.14 The intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0267] like Fig.14As shown, the intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to the program stored in a read-only memory (ROM: ReadOnly Memory) 1002 or the program loaded from a storage device 1003 to a random access memory (RAM: RandomAccess Memory) 1004. In RAM1004, various programs and data required for the operation of the intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 1003 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 1009. The communication device 1009 can allow the intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning to communicate wirelessly or wired with other devices to exchange data. Although the figure shows an intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems may be implemented or have instead.

[0268] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0269] The intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning provided by the present application adopts the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning in the above-mentioned embodiment, which can solve the technical problem of how to intelligently optimize roadbed structure parameters more efficiently and accurately. Compared with the prior art, the beneficial effects of the intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning provided by the present application are the same as the beneficial effects of the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning provided by the above-mentioned embodiment, and the other technical features of the intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0270] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0271] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0272] The present application provides a computer-readable storage medium having computer-readable program instructions (i.e., computer programs) stored thereon, and the computer-readable program instructions are used to execute the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning in the above-mentioned embodiment.

[0273] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0274] The above-mentioned computer-readable storage medium may be included in the intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning; or it may exist independently without being assembled into the intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning.

[0275] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning, the intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning: obtains a set of roadbed engineering parameters to be optimized, a roadbed engineering parameter optimization model and an adaptive hyper-heuristic algorithm; performs population genetics and reverse learning to generate a population based on the set of roadbed engineering parameters to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, and determines the target population individual information; performs correlation analysis based on the set of roadbed engineering parameters to be optimized, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm and the target population individual information to adjust the search direction and search step, determine the target optimization population information, and complete the intelligent optimization of roadbed structure parameters based on the target optimization population information.

[0276] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0277] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0278] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0279] The readable storage medium provided by the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning, and can solve the technical problem of how to intelligently optimize roadbed structure parameters more efficiently and accurately. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present application are the same as the beneficial effects of the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning provided by the above-mentioned embodiment, and will not be repeated here.

[0280] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. An intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning, characterized in that: The method includes: Obtaining the parameter set to be optimized for the roadbed engineering, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm; Based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, population genetics and reverse learning are performed to generate a population, and individual information of the target population is determined; Based on the set of parameters to be optimized for the roadbed engineering, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm and the target population individual information, correlation analysis is performed to adjust the search direction and search step, determine the target optimization population information, and complete the intelligent optimization of roadbed structure parameters based on the target optimization population information.

2. The method according to claim 1, characterized in that The steps of obtaining the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm include: Obtaining a combination of roadbed parameters, geotechnical calculation functions, an allowable set of roadbed engineering parameters, high-level strategies, and low-level operators, wherein the combination of roadbed parameters includes wall height, wall top width, bottom plate thickness, wall toe overhang length, wall heel overhang length, distance between the front edge of the tenon and the toe of the wall, tenon height, and tenon width; Determine a parameter set and an objective function to be optimized for the roadbed project based on the wall height, the wall top width, the bottom plate thickness, the wall toe overhang length, the wall heel overhang length, the distance between the front edge of the tenon and the wall toe, the tenon height and the tenon width; Establishing a roadbed engineering parameter optimization model based on the roadbed engineering parameter set to be optimized, the objective function, the geotechnical calculation function and the roadbed engineering parameter allowable set; An adaptive hyper-heuristic algorithm is determined based on the high-level strategy and the low-level operator.

3. The method according to claim 1, characterized in that The step of generating a population by performing population genetics and reverse learning based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, and determining the target population individual information comprises: Get the greedy learning mechanism, random initial population, random factors and mutation perturbations; The random initial population is screened by a geotechnical calculation program, the engineering cost is calculated and sorted, and the random initial population is reversely learned based on the adaptive hyper-heuristic algorithm to expand the population search space and determine the expanded population; Screening the extended population based on the greedy learning mechanism to determine an iterative population; Improving local search capability based on the random factor and the mutation disturbance, and determining a crossover operator and a mutation operator; Based on the adaptive hyper-heuristic algorithm, the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the iterative population, the crossover operator and the mutation operator, population genetics is performed to select excellent individuals in the population, and repeated population genetics and reverse learning iterative training are triggered to determine the target population individual information.

4. The method according to claim 3, characterized in that The step of performing population genetics to select excellent individuals based on the adaptive hyper-heuristic algorithm, the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the iterative population, the crossover operator and the mutation operator, and triggering repeated population genetics and reverse learning iterative training to determine the target population individual information includes: Determining a current population based on the adaptive hyper-heuristic algorithm, the roadbed engineering parameter set to be optimized, and the roadbed engineering parameter optimization model; Performing fitness evaluation based on the iterative population and the current population to determine the population cost; Selecting excellent individuals of the population based on the population cost as the population parent set; Performing population genetics based on the crossover operator and the population parent set to generate an excellent population, and determining an updated population; The target population individual information is obtained based on the mutation operator and the update population.

5. The method according to claim 1, characterized in that The step of adjusting the search direction and the search step length by performing correlation analysis based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm and the target population individual information to determine the target optimized population information comprises: Obtain target individual component values, target function values, target population parameter means, population standard deviations, and necessary step length sets; Calculate the search direction based on the target population individual information and the objective function in the roadbed engineering parameter optimization model; Calculate a Pearson correlation coefficient set based on the target individual component value, the target function value, the target population parameter mean and the population standard deviation; Calculate the search step length based on the roadbed engineering parameter set to be optimized, the Pearson correlation coefficient set and the necessary step length set; Adjust the search direction and the search step length based on the roadbed engineering parameter set to be optimized and the Pearson correlation coefficient set, and determine the search optimization direction and step length optimization vector; Determining an adaptive operator based on the search optimization direction and the step size optimization vector; Based on the adaptive operator and the target population individual information, the step gradient adaptively updates the population to obtain the target optimized population information.

6. The method according to claim 5, characterized in that The step of adjusting the search direction and the search step length based on the roadbed engineering parameter set to be optimized and the Pearson correlation coefficient set, and determining the search optimization direction and step length optimization vector comprises: Obtain the value range of roadbed engineering parameters and necessary iteration step length; Based on the value range of the roadbed engineering parameter and the necessary iteration step, the roadbed engineering parameter set to be optimized and the search step are selected to determine the roadbed engineering optimization parameter set; The search direction and the search step length are adjusted based on the roadbed engineering optimization parameter set and the Pearson correlation coefficient set to obtain a search optimization direction and a step length optimization vector.

7. The method according to claim 1, characterized in that The step of completing intelligent optimization of roadbed structure parameters based on the target optimization population information includes: Obtain limit iteration number and convergence conditions; When the target optimized population information does not meet the convergence condition and does not meet the limited number of iterations, trigger repeated population training and update the target optimized population information; When the target optimized population information satisfies the convergence condition, the target optimized population information is output.

8. An intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning, characterized in that: The device comprises: An acquisition module is used to acquire a set of parameters to be optimized for a roadbed project, a roadbed project parameter optimization model, and an adaptive hyper-heuristic algorithm; A processing module, used for generating a population by performing population genetics and reverse learning based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model and the adaptive hyper-heuristic algorithm, and determining individual information of the target population; The execution module is used to adjust the search direction and search step size by performing correlation analysis based on the roadbed engineering parameter set to be optimized, the roadbed engineering parameter optimization model, the adaptive hyper-heuristic algorithm and the target population individual information, determine the target optimization population information, and complete the intelligent optimization of the roadbed structure parameters based on the target optimization population information.

9. An intelligent optimization device for roadbed structure parameters integrating population genetics and reverse learning, characterized in that: The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the intelligent optimization method for roadbed structure parameters integrating population genetics and reverse learning as described in any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the roadbed structure parameter intelligent optimization method integrating population genetics and reverse learning as described in any one of claims 1 to 7 are implemented.

Images