Research On Construction Project Scheduling Optimization Problem Based On Network Planning

As the core content of project management,Project scheduling is an important way for enterprises to optimize the production process,promote production efficiency,shorten project duration,and achieve project economic benefits.At present,construction project is developing towards the trend of large-scale,complexity and refinement of division of labors,unreasonable project scheduling can easily cause a series of problems such as time delay,budget overspending and so on,and even leads to project failure.Thus,how to effectively solve the key problems in construction project scheduling and help project managers to formulate scientific and reasonable scheduling scheme has become an important problem to be solved in construction project management.Construction project has the characteristics of large-scale,complex logic relations,and multi resources consumptions,also there are competitions for the limited resources among multi single projects,which increases the complexity of project scheduling,and puts forward higher request for the theoretical model,solving algorithm and practical engineering of the traditional project scheduling problems.In view of this,combined with the characteristics of the construction project,the dissertation takes the project which contains multi single projects and consumes multi resources as the object,and comprehensively applicates the theory and method of project management,operations research and artificial intelligence algorithm to systematically study the construction project scheduling problems.According to different engineering requirements,such as the resources constrained conditions,the different number of the activity execution modes and optimization targets,the dissertation successively studies resource leveling problem without resource constraints,single-mode resource-constrained project scheduling problem,multi-mode resource-constrained project scheduling problem,and multi-mode resource-constrained project scheduling problem with multi-objectives of construction project.For each problem,the dissertation establishes the corresponding mathematical model,because the problem belongs to the NP-Hard problem,thus the dissertation proposes a quick and efficient meta-heuristic algorithm to solve each problem.The proposed algorithms were verified by the cases in project scheduling problem library and specific construction projects.The innovation of the dissertation includes the following four aspects:(1)The dissertation establishes the model of the construction project resource leveling problem,and proposes a hybrid evolutionary algorithm for solving the problem.The model aims at minimizing the degree of multi resource leveling among multiple single projects.In order to make the research feasible,we propose the method of transforming network graph by adding auxiliary work and determine the weight of different resource by analytic hierarchy process(AHP)to realize the comprehensive evaluation of the multi resource leveling degree.The proposed hybrid evolutionary algorithm refers the idea of the simulated annealing algorithm to introduce a linear cooling scheme and metropolis criterion to control the search space of the genetic algorithm,it integrates the global search ability of genetic algorithms with fine tuning capabilities of the simulated annealing technique to achieve an efficient algorithm for the problem.Also the algorithm proposes a dynamic invalid solution repair mechanism based on the float time to avoid the infeasible solution reducing the search efficiency.The effectiveness of the algorithm is verified by a case of construction project.(2)The dissertation establishes the model of the single-mode resource-constrained construction project scheduling problem,and proposes a modified genetic algorithm for solving the problem.The model aims to achieve optimal project scheduling between multiple single projects,which is an extension of classic resource constrained project scheduling problem in the field of Engineering.According to the characteristic of the RCPSP,the proposed algorithm randomly rearranges the sequence of activities with the same start time(finish time)to enhance the diversity of the population,embeds elitist strategy to enhance the convergence rate,modifies crossover order to improve the algorithm's global search ability,and designs a mutation operator which is controlled by a linear decreasing probability to enhance the algorithm's local search ability.The proposed algorithm is tested on 1560 cases of J30,J60,and J120 problems in an international project scheduling problems library and specific construction project,compared with other excellent algorithms in the literature,the validity of the algorithm is verified.(3)The dissertation establishes the model of the multi-mode resource-constrained construction project scheduling problem with fuzzy project duration,and proposes an optimization algorithm to deal with the fuzzy project schedule.According to the engineering practice,the dissertation further considers the uncertainty impact of environmental changes on activity duration and project duration on the basis of multi-mode resource-constrained project scheduling problem,and express the uncertain project duration and activity duration by fuzzy numbers.It modifies the proposed genetic algorithm for single mode problem to solve the problem.The algorithm encodes based on activity list and mode list representation,and modifies parallel schedule generation scheme based on fuzzy mathematics theory to produce fuzzy project schedule.The performance of the algorithm,is verified on cases of J20 problem in PSPLIB and specific construction project.The research shows that it provides a new way to solve the uncertain construction project scheduling problem.(4)The dissertation establishes the model of multi objective resource constrained construction project scheduling problem,and proposes a multi-objective genetic algorithm based on the strategy of Pareto optimal solution.The model considers multi modes of each activity,under the conditions of renewable resource constraints and non-renewable resource constraints,it optimizes time and cost simultaneously.The proposed algorithm uses non-dominated ranking and crowded distance to distribute virtual fitness,which is useful to improve the efficiency of constructing a non-dominated set,also the selection operator based on individual distance matching strategy is proposed to guide global search process,and improve the algorithm's search efficiency.The validity of the algorithm is verified by a construction project.