Research On Project Scheduling Methods Under Uncertainty And Its Implementation

Effective project management is an important approach to improve the competitiveness of enterprise, and the core point is the project scheduling and control. Previous works mostly focused on scheduling problems based on deterministic mathematical models with an assumption that the scheduling environment is ideal, thus there is large gap between theory and practice of project scheduling. There are many kinds of uncertainties in the practical project execution, such as variant task duration, due date changes. Considering the influence of uncertainties when generating the schedule is the key problem to bridge the gap between theory and practice. In the dynamic and uncertain environments, scheduling becomes more complicate than the deterministic and static one, which brings many difficulties to directly use the methods for the static problem.The aim of project scheduling is to guide the actual production process. The production mode of product development project is the single-piece-small-batches. The project scheduling of product development project concerns with process management, and should by no means be separated from the business process and its managerial factors. The realization of scheduling should base on rational and effective project management. Thus it is necessary to analyze and model the business process and project management to effectively apply the scheduling approaches and to improve the executive ability of product development project.The uncertainties which disturb the project execution are boiled down to the time uncertainty, the project scheduling under time uncertainty is studied. The main research work lies in five aspects as follows.1) The determined resource constrained project scheduling is studied. A heuristic iterated local search procedure is presented. The measure is designed to avoid useless search action. The double-shift is used to improve the ability of local search, while two perturbation procedures are designed to overcome local optimality. The results show that the presented ILS procedure can effectively improve the scheduling quality. As the baseline schedule, the result of RCPSP is the basic of research on scheduling under uncertainty.2) The resource constrained project scheduling with stochastic duration is investigated with the objective of generating the optimal scheduling sequence. A heuristic method which is designed using activity based priority policy is presented. The results tested by sampling duration vectors show that the method can improve the scheduling quality. The earliest start policy and resource based priority policy are combined. By means of combination the earliest start policy and the resource based priority policy, and two resource arc allocation methods, a resource arcs random sampling method and a non-standard GA approach are presented. The scheduling quality can be further improved by these two methods.3) The proactive schedule method is proposed under the objective of minimizing the solution robustness. Proactive scheduling focuses at absorbing future partial known disruptions. An integrated optimization process is designed based on the baseline schedule without putting the duration uncertainty into consideration. The optimization of baseline schedule and the time buffer allocation and optimization are combined into one process; the buffered serial scheduling generation scheme is presented. In order to decrease the complexity of intermediate process, a surrogate objective function is used to simplify the evaluation of robustness. The effectiveness of the presented method is tested by simulation.4) The reactive scheduling method is studied. The reactive schedule procedure denotes to how to react when unknown disruption occurs during the project execution. For the reactive scheduling problem, two methods, a particle swarm optimization based meta-heuristic method and a heuristic search approach, are presented combined with reactive policy and scheduling generation scheme. Four local search procedures are designed to improve the search quality. The simulation experiments show that the presented approaches can satisfy the simple and prompt requirements of reactive procedure.5) Aiming at the business process and its management factors in product development project, the technical and managerial factors are integrated. Task management is standardized by the project template. The different roles of risk management for the project plan, scheduling and control phases are pointed out. The extended calculation method for task priority is proposed in consideration of rework risk of design task. For the project management and scheduling, the business process, scheduling use case and data model, et al., are analyzed and modeled by object-oriented method. Finally, parts of project management and scheduling functions are developed and implemented.