Research On The Improved Genetic Algorithm Of Multi-mode Resource-constrained Project Scheduling Problem

In the 1950s,the development of market economy began to enter the "fast track",which was the beginning of the economic competition.Today's market is becoming more and more intense.Low-cost,short-duration and high-quality has become the target of the project management instead of the unilateral requirement.Resource-constrained project scheduling problem has became one of the hot topic in today's modern project management scheduling,which requires achieving a certain optimal goal by arranging the start time and the required resource for all activities under the conditions of both project temporal and resource constraint.However the current domestic study of this problem is mostly based on the assumption of each activity only has one fixed execution model,which is not conform to the actual.In reality,activities mostly have several alternative execution modes.Thus,this paper is the study on the multi-mode resource-constrained project scheduling problem.Multi-mode resource-constrained project scheduling problem has widely propagated in manufacturing,the engineering industry and the IT industry,which is a typical NP-hard problem in project scheduling problem.Compared to the traditional RCPSP,the characteristics of MRCPSP are very outstanding,which has many optional execution modes and each activity execution mode includes different duration and resources required.The study of this issue can not only to extend the multi-faceted solution of modern project,but also can save resource and funds that meets the current requirement of the resource-saving social construction.The main content of this paper are based on the improved genetic algorithms multi-mode resource-constrained project scheduling problem study.Firstly,introduce the related theoretical knowledge and the research status at home and abroad of resource-constrained project scheduling problem,genetic algorithm and PSPLIB.Then according to the characteristics of MRCPSP build the mathematical model with the project duration-optimal goal.In order to solve the above mathematical model,this paper puts forward a new intelligence algorithm based on the genetic algorithm,that called Level Project Scheduling Optimization based on Genetic Algorithm.The most changing in LPSGA algorithm is encoding operation,which divides the project activities into levels,and then determines the number of levels of each activity,and always put the activities with small level number at the front of activities with bigger level number,to ensure the effectiveness of individual coding.The decoding operation,crossover and mutation operation are the key parts of the LPSGA algorithm,which meet the characteristics of MRCPSP.In the respect of validating performance of the proposed the improved GA algorithm,we select the J18,J20 test set in multi-mode from PSPLIB as benchmark base,and the experiment results prove the effectiveness and veracity of the improved GA algorithm.Besides,we apply this algorithm to the actual case,which proves the practical value of the proved algorithm.