Research On The Resource-constrained Maintenance Tasks Network Scheduling Problem

The equipment maintenance system is a collection of a series technical management activities which ensure that the equipment can be maintained to its normal state.If the maintenance activities can be scheduling reasonably,the enterprises will solve problems quickly and avoid accidents.At the same time,jobs can be completed on time so that the enterprises keep working efficiently and improve comprehensive benefits.In the maintenance system,how to optimize the scheduling of resource for maintenance tasks is a key issue.The maintenance tasks network scheduling problem is a resource-constrained project scheduling problem(RCPSP).Because of the complexity of the real condition,the algorithms of solving classic RCPSP are no longer applicable for the maintenance task network.So according to the actual needs of an enterprise,we propose a resource-constrained maintenance tasks network scheduling problem model to solve practical situation.On the basis of RCPSP,we introduce the concepts of sites,maintenance personnel and a series of resource constraints.The objective function of the model is minimizing the maximum completion time.Comparing with the exact algorithm,the heuristic algorithm can solve the large-scale problem in an acceptable time,and obtain a better solution,so we use the heuristic algorithm to solve resource-constrained maintenance tasks network scheduling problem.First,a serial-schedule-generation-scheme constructive heuristic algorithm with four priority rules is proposed to generate an initial solution.The priority rules are used to select the site,operation,renewable resource and maintenance personnel.In order to optimize the choice of resource,we use the max-min ant system in which the updating of the pheromone is limited to optimize the sites' choice.Then we studied genetic algorithm(GA)and particle swarm optimization(PSO),and proposes a hybrid optimization algorithm based on PSO and GA to optimize the choice of sites and maintenance personal.In the algorithm,the genetic factors concluding selection,crossover and mutation are applied to the particle update rules.The results of experiments show that both of the optimization algorithms have a well solution.By comparing the results of the two algorithm and the evaluation in the simulation software,the hybrid optimization algorithm based on PSO and GA have a better optimization effect.It has been proved that the model of resource-constrained maintenance tasks network scheduling problem that proposed in this thesis is reasonable.The algorithm we designed can obtain correct and better results,which can guide the maintenance operations in real station.