Research On Intelligent Optimization Algorithms For Resource-Constrained Multi-Project Scheduling Problem With Planned Resource-Unavailability

Resource-constrained multi-project scheduling problem(RCMPSP)with resource-unavailability is an extension of the classic resource-constrained multi-project scheduling problem(RCMPSP),and it is common in practice.Since such problems are usually complicated and diverse,it is difficult to establish a unified problem model to solve,and there lack researches in literature.In this paper,based on a real producing scenario,a model is established,and optimization algorithms are researched.The main contribution of this work is as follows.First,a resource-constrained multi-project scheduling problem with Planned Resource-Unavailability model(RCMPSP-PRU)is established to minimize the makespan of the project.Different from the traditional RCMPSP,RCMPSP-PRU introduces some new concepts,such as site,movable resource,unmovable resource and planned resource-unavailability,and accompanied with a variety of complex constraints.Second,the Serial Schedule Generation Scheme is improved and the scheduling priority rules are designed according to the characteristics of the problem,so as to an improved serial schedule generation algorithm based on priority rules,named ISSPR,is proposed.The experiments show that ISSPR can build a reasonable and feasible scheduling plan quickly.Further,two single-factor optimization algorithms,GAJTO and TSPRU,are proposed,with respect to the sequence of job transfering and the planned unavailability of resources respectively,which have significant influence on the performance of ISSPR.The GAJTO is a genetic algorithm,and the TSPRU is a tabu search algorithm.The experiments show that both GAJTO and TSPRU have obvious optimization on the solution achieved by ISSPR.However,the GAJTO has the problems of slow convergence and quite time-consuming.Finally,in order to solve the time-consuming problem of GAJTO,a hybrid algorithm of genetic algorithm and particle swarm optimization,named JTO-PSOGA,is proposed.The experiments show that JTO-PSOGA can significantly improve the solving efficiency while not worsen the optimization ability of GAJTO,and usually even better performance.Then,based on the JTO-PSOGA and TSPRU,the algorithm JP-PSOGATS,which is hybrided the genetic algorithm,particle swarm optimization and tabu search,is designed.In JP-PSOGATS,the hybrid algorithm of GA and PSO are used to enhance the job transferring sequence,and the TS is used to deal with the resource unavailability.The experiments show that JP-PSOGATS can achieve better results than any other algorithms in this paper,though it needs more solving time.Considering the characteristics of RCMPSP-PRU,JP-PSOGATS has the best performance with balanced solving quality and efficiency.