Multi-Mode Resource Constrained Project Scheduling Using Differential Evolution Algorithm

摘要

Project scheduling is a tool that manages the work and resources associated with delivering a project on time. Project scheduling is important to organize, keep track of the finished and in-progress tasks and manage the quality of work delivered. However, many problems arise during project scheduling. Minimizing project duration is the primary objective. Project cost is also a critical matter, but there will always be a trade off between project time and cost (Ghoddousiet et al., 2013), so scheduling activities can be challenging due to precedence activities, resources, and execution modes. Schedule reduction is heavily dependent on the availability of resources (Zhuo et al., 2013).;There have been several methods used to solve the project scheduling problem. This dissertation will focus on finding the optimal solution with minimum makespan at lowest possible cost. Schedules should help manage the project and not give a general estimate of the project duration. It is important to have realistic time estimates and resources to give accurate schedules. Generally, project scheduling problems are challenging from a computational point of view (Brucker et al., 1999).;This dissertation applies the differential evolution algorithm (DEA) to multi mode, multi resource constrained project scheduling problems. DEA was applied to a common 14- task network through different scenarios, which includes Multi Mode Single Non Renewable Resource Constrained Project Scheduling Problem (MMSNR) and Multi Mode Multiple Non Renewable Resource Constrained Project Scheduling Problem (MMMNR). DEA was also applied when each scenario was faced with a weekly constraint and when cost and time contingencies such as budget drops or change in expected project completion times interfere with the initial project scheduling plan. A benchmark problem was also presented to compare the DEA results with other optimization techniques such as a genetic algorithm (GA), a particle swarm optimization (PSO) and ant colony optimization (ACO). The results indicated that our DEA performs at least as good as these techniques as far as the project time is concerned and outperforms them in computational times and success rates. Finally, a pareto frontier was investigated, resulting in optimal solutions for a multi objective problem focusing on the tradeoff of the constrained set of parameters.