An intelligent optimization method for housing construction scheduling scheme

Abstract

The present invention discloses an intelligent optimization method for a housing construction scheduling scheme, and belongs to the technical field of construction scheduling optimization. The method comprises the following steps: establishing a multi-objective particle swarm construction scheduling optimization model with the shortest construction period, the lowest construction cost and the highest construction quality as the objectives; adopting the Pareto principle, an external storage library and an adaptive grid to select individual historical optimal solutions and global optimal solutions; establishing a construction scheduling scheme to be optimized, solving the priority constraint problem, and improving the operability and universality of the above model in solving the intelligent optimization problem of housing construction scheduling schemes. The present invention has the advantages of high feasibility and strong ability to search for the optimal solution, and can provide decision makers with a reasonable housing construction scheduling optimization scheme to minimize the construction period and construction cost while ensuring the construction quality.

Description

Technical Field

[0001] The invention relates to the field of construction engineering scheduling, and in particular to an intelligent optimization method for a housing construction scheduling scheme. Background Art

[0002] Construction scheduling optimization has always been a very important research field in the construction industry. Its importance is not only reflected in the model method level, but also in the practical application level. When formulating a construction scheduling plan, how to shorten the construction period, reduce costs, reduce resource consumption, and improve construction quality has always been an important challenge faced by relevant practitioners.

[0003] At present, the existing optimization methods applied to the construction scheduling of housing buildings cannot fully consider the differences between different construction projects, and the practicality and operability of the methods are low. Therefore, there is an urgent need for an easy-to-implement and high-precision intelligent optimization method for the construction scheduling of housing buildings. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent optimization method for a house construction scheduling scheme to solve the problems existing in the prior art.

[0005] The technical solution adopted to achieve the purpose of the present invention is as follows: a method for intelligent optimization of a building construction scheduling scheme, comprising the following steps:

[0006] 1) Establishing a multi-objective particle swarm housing construction scheduling optimization model with the shortest construction period, the lowest construction cost and the highest construction quality as the optimization objectives, and determining the total number of housing construction scheduling schemes in the model, that is, determining the particle swarm scale;

[0007] 2) Establish a scheme to be optimized for house construction scheduling, determine the parameter variables and constraints required by the model, and improve its universality in solving the optimization problem of house construction scheduling;

[0008] 3) Setting the values ​​of the learning factor, inertia weight and mutation operator in the model, setting the upper limit of the number of construction scheduling solutions required and the maximum number of iterations of the model, the upper limit of the number of construction scheduling solutions required includes the maximum allowable value of the external repository;

[0009] 4) Setting the value range of resource options and selection trends for each construction activity in each scheme, that is, determining the value range of particle position and velocity, and randomly initializing the values ​​of resource options and selection trends for the construction activity within the value range, that is, initializing the position and velocity of the particle;

[0010] 5) Calculating the construction period, construction cost and construction quality of each scheme according to the objective function, that is, calculating the fitness value;

[0011] 6) Evaluate each plan individually and select the best combination of construction activity resource options as the individual historical optimal plan according to the Pareto dominance principle, that is, update the individual historical optimal plan;

[0012] 7) Evaluate all the solutions, and according to the Pareto dominance principle, select the non-inferior solutions from all the solutions, store them in an external repository as archived solutions, and perform a second round of screening in the external repository to remove the inferior solutions from the archived solutions;

[0013] 8) Determine whether the number of archived solutions exceeds the maximum allowable value of the external repository. If so, filter according to the adaptive grid density and remove redundant solutions. Otherwise, proceed to the next step.

[0014] 9) Determine whether the loop execution reaches the above maximum number of iterations or the solution set converges. If so, output the archived solution, that is, output the non-inferior solution set in the external repository solution. Otherwise, add 1 to the number of iterations and then execute step 10);

[0015] 10) Using the roulette wheel selection method to select the best combination of construction activity resource options from the archived plans as the global optimal plan, that is, updating the global optimal plan;

[0016] 11) Randomly initialize the schemes within the variation range, i.e., perform particle variation operation;

[0017] 12) Calculate the resource options and selection trends of each construction activity in each scheme according to the update formula of the multi-objective particle swarm optimization algorithm, that is, update the speed and position of the particles, and return to execute step 5).

[0018] Furthermore, in step 1), the construction period, construction cost and construction quality are taken as optimization targets at the same time, and the optimization objective function is:

[0019] Minimize construction period PT min :

[0020]

[0021] Where: i represents the sequence number of the construction activity, n represents the sequence number of the resource option, T i，n It represents the construction duration of construction activity i on the critical path when resource option n is selected;

[0022] Minimize construction costs PC min :

[0023]

[0024] Where: C i，n represents the construction cost consumed when construction activity i selects resource option n; maximizes construction quality PQ max :

[0025]

[0026] Where: w i，n represents the weight of construction activity i in the overall quality performance of the project, Q i，n It represents the construction quality performance when construction activity i selects resource option n.

[0027] Furthermore, the construction scheduling scheme to be optimized in step 2) is a built-in table of the model, which reflects the differences between different housing construction projects through the following three input parameters: building area A 10,000 m 2 , the number of equipment lines B and the housing construction cost index R yuan / m in the project area 2 After determining the key construction activities of the building and their priority constraints, the construction period, construction cost and construction quality of each construction activity are quantified to form a construction scheduling optimization plan.

[0028] Further, step 10) adopts a method of introducing an external repository to determine a global optimal solution, and the external repository includes an archive controller and a grid;

[0029] Further, the archive controller is used to determine whether to add the solution NS to the grid of the external repository, including the following steps:

[0030] If the external repository is empty, then the solution NS is stored; if the external repository already has a solution S1 and S1 dominates NS, then the solution S1 is still stored; if the external repository already has a solution S1 and NS dominates S1, then the solution NS is stored and the solution S1 is removed; if the external repository already has a solution S1 and NS and S1 have no dominating relationship, then the solutions NS and S1 are stored.

[0031] Furthermore, the grid is an objective function space formed by a number of hypercubes, and when generating a uniformly distributed solution set, the following steps are included:

[0032] 10-1) Determine the boundary of the grid: At the tth iteration, the grid boundary is (minPT t , maxPT t )、(minPC t , maxPC t ) and (minPQ t , maxPQ t ), the minimum and maximum fitness values ​​of the particles stored in the external repository calculated according to the objective functions of construction period PT, construction cost PC and construction quality PQ;

[0033] 10-2) Calculate the modulus of each grid:

[0034]

[0035]

[0036]

[0037] Where: ΔPT t is the modulus of the grid in the PT direction of the objective function at the tth iteration; ΔPC t is the modulus of the grid in the PC direction of the objective function at the tth iteration; ΔPQ t is the modulus of the grid in the PQ direction of the objective function at the tth iteration; D is the divisor, which is used to determine the number of hypercubes in the grid;

[0038] 10-3) Calculate and determine the grid numbers of all particles stored in the external repository,

[0039]

[0040] in, is the fitness value calculated by the PT objective function of particle u in the grid at the tth iteration, and Int is a function that rounds down to an integer;

[0041] 10-4) Determine the number of particles in each grid, i.e., the particle density;

[0042] 10-5) Use the roulette method to select the global optimal solution.

[0043] Furthermore, in step 12), the particle updates its speed according to the individual historical optimal solution and the global optimal solution, and then updates its position according to the current speed. The speed update formula is:

[0044] v k (t+1)=ω·v k (t)+c1·r1(P k (t)-x k (t))+c2·r2(R g (t)-x k (t)) (7)

[0045] Displacement update formula:

[0046] x k (t+1)=x k (t)+v k (t+1) (8)

[0047] Where: k = 1, 2...K, K is the total number of particles in the particle swarm; t = 1, 2...E, E is the limit number of iterations; v k(t) is the velocity of the kth particle at the tth iteration; x k (t) is the position of the kth particle at the tth iteration; P k (t) is the individual historical optimal solution of the particle selected by the dominance relationship; R g (t) is the global optimal solution of the particle selected from the external repository at the tth iteration; ω is the inertia weight, a non-negative number used to control the influence of the previous speed on the current speed; r1 and r2 are random numbers with a value range of [0, 1]; c1 and c2 are acceleration constants, i.e., learning factors;

[0048] After the particles are initialized and updated, the particle position and velocity values ​​that are out of range are adjusted to the limit values.

[0049] Furthermore, in step 11), the entire population is divided into three parts of equal size, and different mutation operators are used for the three parts; the particles of the first sub-part will not mutate at all; the particles of the second sub-part will adopt uniform mutation, that is, the particles that meet the mutation requirements within the range are randomly initialized, and the mutation range does not change with the number of iterations; the particles of the third sub-part will adopt non-uniform mutation, that is, the particles that meet the mutation requirements within the range are randomly initialized, and the mutation range decreases with the number of iterations.

[0050] The beneficial effects of the present invention are:

[0051] 1. The construction scheduling optimization model constructed by the present invention is suitable for solving multi-objective construction scheduling optimization problems;

[0052] 2. The housing construction scheduling optimization scheme established by the present invention effectively improves the universality and operability of the above-mentioned model in solving the housing construction scheduling optimization problem;

[0053] 3. The present invention adopts external storage, adaptive grid method and roulette method to improve the convergence of the housing construction scheduling optimization model;

[0054] 4. The present invention adopts three particle mutation operations to effectively prevent the housing construction scheduling optimization model from falling into local convergence;

[0055] 5. The present invention can provide decision makers with a construction scheduling solution that meets the optimization objectives, so as to minimize the construction period and construction cost, while maximizing the construction quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of the process of the present invention;

[0057] Figure 2 Optimize the target diagram for construction scheduling;

[0058] Figure 3 This is a schematic diagram of the archive controller's function;

[0059] Figure 4 Schematic diagram of optimization iteration results. DETAILED DESCRIPTION

[0060] The present invention is further described below in conjunction with the embodiments, but it should not be understood that the above subject matter of the present invention is limited to the following embodiments. Without departing from the above technical ideas of the present invention, various substitutions and changes are made according to the common technical knowledge and customary means in the art, which should all be included in the protection scope of the present invention.

[0061] Embodiment 1:

[0062] See also Figure 1 This embodiment discloses an intelligent optimization method for a housing construction scheduling scheme, comprising the following steps:

[0063] 1) A multi-objective particle swarm housing construction scheduling optimization model is established with the shortest construction period, the lowest construction cost and the highest construction quality as the optimization objectives, and the total number of housing construction scheduling schemes in the model is determined, that is, the particle swarm scale is determined, which is set to 200 in this embodiment; wherein, the construction period, construction cost and construction quality are simultaneously taken as the optimization objectives, and the optimization objective function is:

[0064] Minimize construction period PT min :

[0065]

[0066] Where: i represents the sequence number of the construction activity, n represents the sequence number of the resource option, T i，n It represents the construction duration of construction activity i on the critical path when resource option n is selected;

[0067] Minimize construction costs PC min :

[0068]

[0069] Where: C i，n represents the construction cost consumed when construction activity i selects resource option n; maximizes construction quality PQ max :

[0070]

[0071] Where: w i，n represents the weight of construction activity i in the overall quality performance of the project, Q i，n represents the construction quality performance when construction activity i selects resource option n; see Figure 2, is a schematic diagram of the construction scheduling optimization target of this embodiment;

[0072] 2) Establish a housing construction scheduling optimization plan, determine the parameter variables and constraints required by the model, and improve its universality in solving the housing construction scheduling optimization problem; wherein, the construction scheduling optimization plan is a built-in table of the model, and the differences between different housing construction projects are reflected through the following three input parameters: Building area A 10,000 m 2 , the number of equipment lines B and the housing construction cost index R yuan / m in the project area 2 After determining the key construction activities of the building and the priority constraints between them, the construction period, construction cost and construction quality of each construction activity are quantified to form a construction scheduling optimization scheme. Table 1 is a construction scheduling optimization scheme provided by an embodiment of the present invention:

[0073] Table 1

[0074]

[0075]

[0076] The data sources in Table 1 are as follows: the construction period is determined according to the National Unified Construction and Installation Project Period Quota (TY01-89-2016) combined with actual project data; the construction cost is determined according to the Chongqing Housing Construction and Decoration Engineering Pricing Quota (CQJZDE-2018) and the Chongqing Construction Project Cost Quota (CQFYDE-2018); it is converted through housing construction cost indicators to make it applicable to housing construction in various places; the construction quality performance is quantified by the expert scoring method and hierarchical analysis method in terms of its scoring and weight.

[0077] 3) Setting the values ​​of the learning factor, inertia weight and mutation operator in the model. In this embodiment, a linearly decreasing inertia weight is adopted, the inertia weight range is 0.4-0.95, the learning factor is 2, and the mutation operator is randomly selected in the range of [0,1]; setting the upper limit of the number of construction scheduling solutions required and the maximum number of iterations of the model. The upper limit of the number of construction scheduling solutions required includes the maximum allowable value of the external repository, which is set to 200 in this embodiment, and the maximum number of iterations is set to 120 times;

[0078] 4) Set the value range of resource options and selection trends of each construction activity in each plan, that is, determine the value range of particle position and speed, and randomly initialize the resource options and selection trend of the construction activity within the value range, that is, initialize the position and speed of the particle; in this embodiment, the resource option combination of each construction activity in the construction scheduling plan is used as the position of each particle, and the value is an integer ranging from 1 to Mi (i=1,...,N), Mi is the number of resource options for construction activity i in Table 1, so the particle position range is limited to [1,Mi], and rounded to an integer. The selection trend of the construction activity, that is, the particle speed range is limited to [1-Mi,Mi-1];

[0079] 5) Calculating the construction period, construction cost and construction quality of each scheme according to the objective function, that is, calculating the fitness value;

[0080] 6) Evaluate each plan individually and select the best combination of construction activity resource options as the individual historical optimal plan according to the Pareto dominance principle, that is, update the individual historical optimal plan;

[0081] 7) Evaluate all the solutions, and according to the Pareto dominance principle, select the non-inferior solutions from all the solutions, store them in an external repository as archived solutions, and perform a second round of screening in the external repository to remove the inferior solutions from the archived solutions;

[0082] 8) Determine whether the number of archived solutions exceeds the maximum allowable value of the external repository. If so, filter according to the adaptive grid density and remove redundant solutions. Otherwise, proceed to the next step.

[0083] 9) Determine whether the loop execution reaches the above maximum number of iterations or the solution set converges. If so, output the archived solution, that is, output the non-inferior solution set in the external repository solution. Otherwise, add 1 to the number of iterations and then execute step 10);

[0084] 10) The roulette wheel selection method is used to select the best combination of construction activity resource options from the archived solutions as the global optimal solution, that is, to update the global optimal solution; wherein, the global optimal solution is determined by introducing an external repository, and the external repository includes an archive controller and a grid; the archive controller is used to determine whether to add the solution NS to the grid of the external repository, Figure 3 The schematic diagram of the archive controller provided in the embodiment of the present invention includes the following steps:

[0085] If the external repository is empty, then the solution NS is stored; if the external repository already has solution S1, and S1 dominates NS, then solution S1 is still stored; if the external repository already has solution S1, and NS dominates S1, then solution NS is stored and solution S1 is removed; if solution S1 already exists in the external repository, and NS and S1 have no dominating relationship, then solutions NS and S1 are stored;

[0086] The grid is an objective function space formed by a number of hypercubes. When generating a uniformly distributed solution set, the following steps are included:

[0087] 10-1) Determine the boundary of the grid: At the tth iteration, the grid boundary is (minPT t , maxPT t )、(minPC t , maxPC t ) and (minPQ t , maxPQ t ), the minimum and maximum fitness values ​​of the particles stored in the external repository calculated according to the objective functions of construction period PT, construction cost PC and construction quality PQ;

[0088] 10-2) Calculate the modulus of each grid:

[0089]

[0090]

[0091]

[0092] Where: ΔPT t is the modulus of the grid in the PT direction of the objective function at the tth iteration; ΔPC t is the modulus of the grid in the PC direction of the objective function at the tth iteration; ΔPQ t is the modulus of the grid in the PQ direction of the objective function at the tth iteration; D is the divisor, which is used to determine the number of hypercubes in the grid;

[0093] 10-3) Calculate and determine the grid numbers of all particles stored in the external repository,

[0094]

[0095] in, is the fitness value calculated by the PT objective function of particle u in the grid at the tth iteration, and Int is a function that rounds down to an integer;

[0096] 10-4) Determine the number of particles in each grid, i.e., the particle density;

[0097] 10-5) Use the roulette method to select the global optimal solution;

[0098] In addition, the roulette wheel method selects the global optimal solution. To ensure the diversity of the non-inferior solution set and explore more space, the lower the particle density in the grid, the greater the probability of being selected; use 10 divided by the number of particles in the unit grid z j As the fitness of the j-th unit grid f(z j )(j=1,2,...,J), J is the number of unit grids containing particles. Calculate the fitness ratio, that is, the ratio of the fitness of the jth unit grid to the sum of all fitness, such as Where p(z j ) is the unit grid z j The fitness ratio, f(z j ) is the unit grid z j Calculate the cumulative probability q j , that is, the sum of the fitness ratios from the first unit grid to the jth unit grid, such as Among them, q j is the unit grid z j The cumulative probability of . Randomly generate o∈[0,1], start judging from q1, if q1>o, then select the unit grid 1; if q1<o, then j+1, judge q2, until q j >o, select the unit grid j. Then randomly generate o∈[0,1] again and start judging from q1, making J selections in total. Determine the unit grid that has been selected the most times (the number may be greater than 1), and randomly select a particle in the determined grid as the global optimal solution.

[0099] 11) Randomly initialize the schemes within the mutation range, i.e., perform particle mutation operation; wherein, the entire population is divided into three parts of equal size, and different mutation operators are used for the three parts; the particles in the first sub-part will not mutate at all; the particles in the second sub-part will adopt uniform mutation, i.e., the particles that meet the mutation requirements within the range are randomly initialized, and the mutation range does not change with the number of iterations; the particles in the third sub-part will adopt non-uniform mutation, i.e., the particles that meet the mutation requirements within the range are randomly initialized, and the mutation range decreases with the number of iterations;

[0100] 12) Calculate the resource options and selection trends of each construction activity in each scheme according to the update formula of the multi-objective particle swarm optimization algorithm, that is, update the speed and position of the particles, and return to execute step 5);

[0101] In step 12), the particle updates its speed according to the individual historical optimal solution and the global optimal solution, and then updates its position according to the current speed. The speed update formula is:

[0102] v k (t+1)=ω·v k (t)+c1·r1(P k (t)-x k (t))+c2·r2(R g (t)-x k (t)) (7)

[0103] Displacement update formula:

[0104] x k (t+1)=x k (t)+v k (t+1) (8)

[0105] Where: k = 1, 2...K, K is the total number of particles in the particle swarm; t = 1, 2...E, E is the limit number of iterations; v k (t) is the velocity of the kth particle at the tth iteration; x k (t) is the position of the kth particle at the tth iteration; P k (t) is the individual historical optimal solution of the particle selected by the dominance relationship; R g (t) is the global optimal solution of the particle selected from the external repository at the tth iteration; ω is the inertia weight, a non-negative number ranging from 0.4 to 0.95, which is used to control the influence of the previous speed on the current speed; r1 and r2 are random numbers ranging from [0, 1]; c1 and c2 are acceleration constants, i.e., learning factors, which are 2;

[0106] After the particles are initialized and updated, the particle position and velocity values ​​that are out of range are adjusted to the limit values.

[0107] In this embodiment, the building area A is 24,500 m 2 , equipment lines B = 4, building construction cost index 1216 yuan / m 2 The construction scheduling optimization results output in step 9 are shown in Figure 4 , Figure 4 A schematic diagram of optimization iteration results provided for an embodiment of the present invention, a total of 163 optimized construction scheduling optimization schemes were obtained, all of which were Pareto optimal, with the shortest construction period of 148 days, the lowest cost of 38,457,722 yuan, and the highest quality score of 7.64 points.

[0108] Decision makers can further select solutions based on the requirements of the project to be optimized for construction period, construction cost and construction quality. Assuming that decision makers need to simultaneously meet the construction period of less than 150 days, construction cost less than 40 million yuan, and construction quality higher than 6.5 points, the resulting construction scheduling optimization solution is shown in Table 2. The numbers under the Pareto optimal solution in Table 2 are the resource options for each construction activity in Table 1.

[0109] Table 2

[0110]

[0111]

[0112] Embodiment 2:

[0113] This embodiment discloses a method for intelligent optimization of a building construction scheduling scheme, comprising the following steps:

[0114] 1) Establishing a multi-objective particle swarm housing construction scheduling optimization model with the shortest construction period, the lowest construction cost and the highest construction quality as the optimization objectives, and determining the total number of housing construction scheduling schemes in the model, that is, determining the particle swarm scale;

[0115] 2) Establish a scheme to be optimized for house construction scheduling, determine the parameter variables and constraints required by the model, and improve its universality in solving the optimization problem of house construction scheduling;

[0116] 3) Setting the values ​​of the learning factor, inertia weight and mutation operator in the model, setting the upper limit of the number of construction scheduling solutions required and the maximum number of iterations of the model, the upper limit of the number of construction scheduling solutions required includes the maximum allowable value of the external repository;

[0117] 4) Setting the value range of resource options and selection trends for each construction activity in each scheme, that is, determining the value range of particle position and velocity, and randomly initializing the values ​​of resource options and selection trends for the construction activity within the value range, that is, initializing the position and velocity of the particle;

[0118] 5) Calculating the construction period, construction cost and construction quality of each scheme according to the objective function, that is, calculating the fitness value;

[0119] 6) Evaluate each plan individually and select the best combination of construction activity resource options as the individual historical optimal plan according to the Pareto dominance principle, that is, update the individual historical optimal plan;

[0120] 7) Evaluate all the solutions, and according to the Pareto dominance principle, select the non-inferior solutions from all the solutions, store them in an external repository as archived solutions, and perform a second round of screening in the external repository to remove the inferior solutions from the archived solutions;

[0121] 8) Determine whether the number of archived solutions exceeds the maximum allowable value of the external repository. If so, filter according to the adaptive grid density and remove redundant solutions. Otherwise, proceed to the next step.

[0122] 9) Determine whether the loop execution reaches the above maximum number of iterations or the solution set converges. If so, output the archived solution, that is, output the non-inferior solution set in the external repository solution. Otherwise, add 1 to the number of iterations and then execute step 10);

[0123] 10) Using the roulette wheel selection method to select the best combination of construction activity resource options from the archived plans as the global optimal plan, that is, updating the global optimal plan;

[0124] 11) Randomly initialize the schemes within the variation range, i.e., perform particle variation operation;

[0125] 12) Calculate the resource options and selection trends of each construction activity in each scheme according to the update formula of the multi-objective particle swarm optimization algorithm, that is, update the speed and position of the particles, and return to execute step 5).

[0126] Embodiment 3:

[0127] The main steps of this embodiment are the same as those of Embodiment 2. Furthermore, in step 1), the construction period, construction cost and construction quality are simultaneously taken as optimization targets, and the optimization objective function is:

[0128] Minimize construction period PT min :

[0129]

[0130] Where: i represents the sequence number of the construction activity, n represents the sequence number of the resource option, T i，n It represents the construction duration of construction activity i on the critical path when resource option n is selected;

[0131] Minimize construction costs PC min :

[0132]

[0133] Where: C i，n represents the construction cost consumed when construction activity i selects resource option n; maximizes construction quality PQ max :

[0134]

[0135] Where: w i，n represents the weight of construction activity i in the overall quality performance of the project, Q i，nIt represents the construction quality performance when construction activity i selects resource option n;

[0136] Embodiment 4:

[0137] The main steps of this embodiment are the same as those of Embodiment 2. Furthermore, the construction scheduling scheme to be optimized in step 2) is a built-in table of the model, and the differences between different housing construction projects are reflected through the following three input parameters: Building area A (ten thousand m2) 2 , the number of equipment lines B and the housing construction cost index R yuan / m in the project area 2 After determining the key construction activities of the building and their priority constraints, the construction period, construction cost and construction quality of each construction activity are quantified to form a construction scheduling optimization plan.

[0138] Embodiment 5:

[0139] The main steps of this embodiment are the same as those of Embodiment 2. Further, in step 10), a method of introducing an external repository is used to determine a global optimal solution, and the external repository includes an archive controller and a grid;

[0140] Embodiment 6:

[0141] The main steps of this embodiment are the same as those of Embodiment 5. Further, the archive controller is used to determine whether to add the scheme NS to the grid of the external repository, including the following steps:

[0142] If the external repository is empty, then the solution NS is stored; if the external repository already has a solution S1 and S1 dominates NS, then the solution S1 is still stored; if the external repository already has a solution S1 and NS dominates S1, then the solution NS is stored and the solution S1 is removed; if the external repository already has a solution S1 and NS and S1 have no dominating relationship, then the solutions NS and S1 are stored.

[0143] Embodiment 7:

[0144] The main steps of this embodiment are the same as those of Embodiment 5. Furthermore, the grid is an objective function space formed by a number of hypercubes. When a uniformly distributed solution set is generated, the following steps are included:

[0145] 10-1) Determine the boundary of the grid: At the tth iteration, the grid boundary is (minPT t , maxPT t )、(minPC t , maxPC t ) and (minPQ t , maxPQ t), the minimum and maximum fitness values ​​of the particles stored in the external repository calculated according to the objective functions of construction period PT, construction cost PC and construction quality PQ;

[0146] 10-2) Calculate the modulus of each grid:

[0147]

[0148]

[0149]

[0150] Where: ΔPT t is the modulus of the grid in the PT direction of the objective function at the tth iteration; ΔPC t is the modulus of the grid in the PC direction of the objective function at the tth iteration; ΔPQ t is the modulus of the grid in the PQ direction of the objective function at the tth iteration; D is the divisor, which is used to determine the number of hypercubes in the grid;

[0151] 10-3) Calculate and determine the grid numbers of all particles stored in the external repository,

[0152]

[0153] in, is the fitness value calculated by the PT objective function of particle u in the grid at the tth iteration, and Int is a function that rounds down to an integer;

[0154] 10-4) Determine the number of particles in each grid, i.e., the particle density;

[0155] 10-5) Use the roulette method to select the global optimal solution.

[0156] Embodiment 8:

[0157] The main steps of this embodiment are the same as those of embodiment 5. Further, in step 12), the particle updates its speed according to the individual historical optimal solution and the global optimal solution, and then updates its position according to the current speed. The speed update formula is:

[0158] v k (t+1)=ω·v k (t)+c1·r1(P k (t)-x k (t))+c2·r2(R g (t)-x k (t)) (7)

[0159] Displacement update formula:

[0160] xk (t+1)=x k (t)+v k (t+1) (8)

[0161] Where: k = 1, 2...K, K is the total number of particles in the particle swarm; t = 1, 2...E, E is the limit number of iterations; v k (t) is the velocity of the kth particle at the tth iteration; x k (t) is the position of the kth particle at the tth iteration; P k (t) is the individual historical optimal solution of the particle selected by the dominance relationship; R g (t) is the global optimal solution of the particle selected from the external repository at the tth iteration; ω is the inertia weight, a non-negative number used to control the influence of the previous speed on the current speed; r1 and r2 are random numbers with a value range of [0, 1]; c1 and c2 are acceleration constants, i.e., learning factors;

[0162] After the particles are initialized and updated, the particle position and velocity values ​​that are out of range are adjusted to the limit values.

[0163] Embodiment 9:

[0164] The main steps of this embodiment are the same as those of Embodiment 2. Furthermore, in step 11), the entire population is divided into three parts of equal size, and different mutation operators are used for the three parts; the particles of the first sub-part will not mutate at all; the particles of the second sub-part will adopt uniform mutation, that is, the particles that meet the mutation requirements within the range are randomly initialized, and the mutation range does not change with the number of iterations; the particles of the third sub-part will adopt non-uniform mutation, that is, the particles that meet the mutation requirements within the range are randomly initialized, and the mutation range decreases with the number of iterations.

Claims

1. A method for intelligent optimization of a housing construction scheduling scheme, characterized in that: The following steps are involved: 1) Establishing a multi-objective particle swarm housing construction scheduling optimization model with the shortest construction period, the lowest construction cost and the highest construction quality as the optimization objectives, and determining the total number of housing construction scheduling schemes in the model as the particle swarm scale; 2) Establish a scheme to be optimized for house construction scheduling, determine the parameter variables and constraints required by the model, and improve its universality in solving the optimization problem of house construction scheduling; 3) Setting the values ​​of the learning factor, inertia weight and mutation operator in the model, setting the upper limit of the number of construction scheduling solutions required and the maximum number of iterations of the model, the upper limit of the number of construction scheduling solutions required includes the maximum allowable value of the external repository; 4) setting the value range of each construction activity resource option and selection trend in each scheme as the value range for determining the particle position and speed, and randomly initializing the resource option and selection trend of the construction activity within the value range as the initialization particle position and speed; 5) Calculate the construction period, construction cost and construction quality of each scheme according to the objective function as the calculation fitness value; 6) Evaluate each plan individually and select the best combination of construction activity resource options as the individual historical optimal plan according to the Pareto dominance principle, as the updated individual historical optimal plan; 7) Evaluate all the solutions, and according to the Pareto dominance principle, select the non-inferior solutions from all the solutions, store them in an external repository as archived solutions, and perform a second round of screening in the external repository to remove the inferior solutions from the archived solutions; 8) Determine whether the number of archived solutions exceeds the maximum allowable value of the external repository. If so, filter according to the adaptive grid density and remove redundant solutions. Otherwise, proceed to the next step. 9) Determine whether the loop execution reaches the above maximum number of iterations or the solution set converges. If so, output the archived solution as the non-inferior solution set in the output external repository solution. Otherwise, add 1 to the number of iterations and execute step 10); 10) Using the roulette wheel selection method to select the best combination of construction activity resource options from the archived plans as the global optimal plan, and as the updated global optimal plan; 11) Randomly initialize the schemes within the mutation range as a particle mutation operation; 12) According to the update formula of the multi-objective particle swarm optimization algorithm, the resource options and selection trends of each construction activity in each scheme are calculated as the speed and position of the updated particles, and then the process returns to step 5).

2. The intelligent optimization method for a housing construction scheduling scheme according to claim 1 is characterized in that: In step 1), construction period, construction cost and construction quality are taken as optimization targets at the same time, and the optimization objective function is: Minimize construction period PT min : Where: i represents the sequence number of the construction activity, n represents the sequence number of the resource option, T i,n It represents the construction duration of construction activity i on the critical path when resource option n is selected; Minimize construction costs PC min : Where: C i,n It represents the construction cost consumed when construction activity i selects resource option n; Maximize construction quality PQ max : Where: w i,n represents the weight of construction activity i in the overall quality performance of the project when resource option n is selected, Q i,n It represents the construction quality performance when construction activity i selects resource option n.

3. The intelligent optimization method for a building construction scheduling scheme according to claim 1 or 2, characterized in that: The construction scheduling optimization scheme described in step 2) is a built-in table of the model, which reflects the differences between different housing construction projects through the following three input parameters: Building area A 10,000 m 2 , the number of equipment lines B and the housing construction cost index R yuan / m in the project area 2 After determining the key construction activities of the building and the priority constraints between them, the construction period, construction cost and construction quality of each construction activity are quantified to form a construction scheduling optimization plan.

4. The intelligent optimization method for a building construction scheduling scheme according to claim 1 or 2 is characterized in that: Step 10) A method of introducing an external repository is used to determine the global optimal solution, where the external repository includes an archive controller and a grid.

5. The intelligent optimization method for a housing construction scheduling scheme according to claim 4 is characterized in that: The archive controller is used to determine whether to add the scheme NS to the grid of the external repository, including the following steps: If the external repository is empty, then the solution NS is stored; if the external repository already has a solution S1 and S1 dominates NS, then the solution S1 is still stored; if the external repository already has a solution S1 and NS dominates S1, then the solution NS is stored and the solution S1 is removed; if the external repository already has a solution S1 and NS and S1 have no dominating relationship, then the solutions NS and S1 are stored.

6. The intelligent optimization method for a housing construction scheduling scheme according to claim 4 is characterized in that: The grid is an objective function space formed by a number of hypercubes. When generating a uniformly distributed solution set, the following steps are included: 10-1) Determine the boundary of the grid: At the tth iteration, the grid boundary is (minPT t ,maxPT t )、(minPC t ,maxPC t ) and (minPQ t ,maxPQ t ), the minimum and maximum fitness values ​​of the particles stored in the external repository calculated according to the objective functions of construction period PT, construction cost PC and construction quality PQ; 10-2) Calculate the modulus of each grid: Where: ΔPT t is the modulus of the grid in the PT direction of the objective function at the tth iteration; ΔPC t is the modulus of the grid in the PC direction of the objective function at the tth iteration; ΔPQ t is the modulus of the grid in the PQ direction of the objective function at the tth iteration; D is the divisor, which is used to determine the number of hypercubes in the grid; 10-3) Calculate and determine the grid numbers of all particles stored in the external repository, in, is the fitness value calculated by the PT objective function of particle u in the grid at the tth iteration; is the fitness value calculated by the PC objective function of particle u in the grid at the tth iteration; is the fitness value calculated by the PQ objective function of particle u in the grid at the tth iteration; Int is a function that rounds down to an integer; 10-4) Determine the number of particles in each grid as the particle density; 10-5) Use the roulette method to select the global optimal solution.

7. The intelligent optimization method for a building construction scheduling scheme according to claim 4 is characterized in that: In step 12), the particle updates its speed according to the individual historical optimal solution and the global optimal solution, and then updates its position according to the current speed. The speed update formula is: v k (t+1)=ω·v k (t)+c1·r1(P k (t)-x k (t))+c2·r2(R g (t)-x k (t)) (7) Displacement update formula: x k (t+1)=x k (t)+v k (t+1) (8) Where: k = 1, 2…K, K is the total number of particles in the particle swarm; t = 1, 2…E, E is the limit number of iterations; v k (t) is the velocity of the kth particle at the tth iteration; x k (t) is the position of the kth particle at the tth iteration; P k (t) is the individual historical optimal solution of the particle selected by the dominance relationship; R g (t) is the global optimal solution of the particle selected from the external repository at the tth iteration; ω is the inertia weight, a non-negative number used to control the influence of the previous velocity on the current velocity; r1 and r2 are random numbers with a value range of [0,1]; c1 and c2 are acceleration constants, which are used as learning factors; After the particles are initialized and updated, the particle position and velocity values ​​that are out of range are adjusted to the limit values.

8. The intelligent optimization method for a building construction scheduling scheme according to claim 1 is characterized in that: In step 11), the entire population is divided into three parts of equal size, and different mutation operators are used for these three parts; the particles in the first sub-part will not mutate at all; the particles in the second sub-part will adopt uniform mutation, and the particles that meet the mutation requirements within this range are randomly initialized, and the mutation range does not change with the number of iterations; the particles in the third sub-part will adopt non-uniform mutation, and the particles that meet the mutation requirements within this range are randomly initialized, and the mutation range decreases with the number of iterations.

Images