Research Of Resource Constrained Project Scheduling Problem Based On Multiobjective Evolutionary Algorithm

Resource-constrained project scheduling problem(RCPSP)is a kind of scheduling problems,which is under the condition of satisfying the timing constraints and the resource constraints,to reasonable arrangement the project task,in order to achieve a certain goal optimal.The existing research on RCPSP is mainly concentrated in the single objective optimization,however RCPSP in reality often contain multiple objectives,therefore the research of multi-objective resource-constrained project scheduling problem(MORCPSP)has important theoretical and practical significance.This paper does some research on the MORCPSP,as follows:Firstly,this paper briefly introduces the concept of multi-objective optimization problem and its solution methods,describes the concepts and mathematical models of the classic RCPSP,the multi-mode RCPSP and the multi-objective multi-mode RCPSP,introduces the intelligent optimization algorithms for solving the RCPSP.Secondly,according to the different multi-agent neighbor structure,combined the agent model and knowledge model,a knowledge-guide multi-objective multi-agent evolutionary algorithm is proposed.The different kinds of neighbor environment of the agent are defined.Then the population evolution process is realized through three operators named competition,orthogonal crossover and knowledge learning.A novel method to find the non-dominated sets is designed and the circular crowded sorting method is adopted to maintain the external archive set.The simulation results of several benchmark functions show knowledge-guide multi-objective multi-agent evolutionary algorithm not only increases the diversity of the population but also improves the convergence of the algorithm.Lastly,the knowledge-guide multi-objective multi-agent evolutionary algorithm is applied to solving the MORCPSP.The algorithm uses the task list and model list combined to complete the coding operation,and the serial scheduling scheme to complete the decoding operation.The paper defines the agent evolutionary operation,as competition,cross and self learning.Through the simulation of several projects,showing that the algorithm is effective to solve the MORCPSP.