Research On Optimization Of Multi-project Network Planning Based On Genetic Algorithm

The optimization of the network plan is very important in project management.The core problem is to make the necessary adjustments according to the network planning's target and to establish a reasonable schedule.so that the whole project cost a minimum of resources,funds and the shortest time period.at the same as soon as possible.It determines the level of project profits.Generally,network planning optimization problems included network planning period optimization problem,resource optimization problem,and cost optimization problem.The goal of period Optimization is to continuously shorten the period to meet the project requirements. Each work in the project can be done by many different methods,and different options have different costs and duration.The goal of cost optimization is to get the shortest project duration and cost least money through the methos of selecting the optimal options from all works. Resource optimization includes the optimization of limited resources-the shortest duration and the optimization of period fixed-resources balanced.Now many people study these three problems.However,the research of them is mainly concentrated in the single project and the one objective problem.There had few researches about Multi-project and multi-objecytive.In this paper,according to the actual situation,the optimization of all kinds of the single-objective optimization problems as well as integrated multi-objective optimization problem of the complex multi-project network planning are studied,some improved and new methods are put forward. For the dynamic network planning optimization problem,has not yet found the research literature about this problem.In this paper,a rolling window technique combined with the genetic algorithm has been put forward and successfully solved the problem.A dynamic storage algorithm base on cross list was designed for multi-project.This algorithm can store any number of projects in the cross list,can dynamically increase or decrease of the project,avoid the combining operation to the multi-project network diagram,and increase the flexibility and efficiency of multi-project scheduling.Using a hybrid genetic algorithm studied the multi-project time optimization problem in the situation of resource constrained.First of all,using a variety of heuristic methods and a combination of randomly methods generated initial population.This method improved the quality of the initial solution and ensured the diversity of the initial solution.Then the crossover operator and mutation operator of genetic algorithm has been improved.Crossover operator improved Partially Mapped Crossover operation,and Mutation operator uses a neighborhood search-based method and can improve search efficiency.The state generation function of simulated annealing algorithm was improved.Genetic algorithms and simulated annealing algorithm were combined to optimize the problem.The final experimental results show that this method is better than a variety of heuristic algorithm and two intelligent algorithms.To the multi-project resource leveling optimization problem,build the math model,and then optimized it with genetic algorithm.Design a highly efficient repair operator which can Carry the resources from the day of using the largest amount of resources to the day of using small amount of resources.This method improved the efficiency of the algorithm.Design a genetic algorithm which combined the methods of project scheduling problem and resources balance optimization problem.Fist,compute the shortest period under resource constraints.Then adjust the resources to reduce the using of resources.This method can optimize the two problems at while.Design a genetic algorithm to optimize the multi-project time and cost optimization problem. There are two types of work in project.One is that the time and cost is a continuous relationship, and the other is the dispersion relations.In this paper,using genetic algorithms solve these two types of issues,at the same time taking into account the time limited for project,extended penalties and incentives for completion ahead of schedule.To the time-cost and resource three objective integrated optimization problem about multi-project,using genetic algorithms successfully solve it.Considering the characteristics of network planning problem itself,a comprehensive optimization model of multi-project was established,while the overall optimization process was divided into two phases.In the first phases,resource-constrained optimization and cost optimization problem were combined and optimized at same time.Because of the different characteristics of these two problems,two different types of chromosomes about these two problems were set,and crossover and mutation operation were run respectively.At same time,these two problems still are interdependent,so it is necessary to evaluate them as a whole.On the basis of optimization of the first stage,resource balance optimization which can reduce use of resources was carried out in the second phase.This method can simultaneously optimize the three objectives.Finally,an example from different points of view proved the correctness and high efficiency of the model.For the dynamic multi-project network planning optimization problem,established the mathematical model and put forward the solution which is based on genetic algorithm optimization method for the first time.The rolling window technology in the field of predictive control was applied to the problem.Based on a hybrid-driven strategy which were composed by event-driven strategy and cycle-driven strategy,a genetic algorithm was designed to dynamically optimize the multi-project network planning problem when all kinds of unexpected events happened in the actual projects.The simulation experiments prove that this method can effectively solve a variety of unexpected problems,so it is feasible and efficient.