The Research On Feedback System Of PERT Based On PSO

Construction projects, especially the state's key investment projects often have such character as enormous scale of investment, long construction period, technical risk and economic risks. And there are often impact by the ecological environment and many other uncertain factors. As result it is often conduced that the consequences of the actual construction of the project have some degree of departure from what people had expected. Schedule control is one important goal of three: the project investment control, schedule control and quality control, which has a direct impact on the other two goals executing. When the schedule of the project was out of control or was substantially delayed, the contractor should not only increase the human power and material to compensate progress, but also have to pay liquidated damages of the delay. What's more, the investor can not benefit from the construction product. It is seriously affected multi-stakeholder. Therefore, it is of great importance to have a reasonable control of project schedule, which derived from accurately prediction of the schedule of the project and estimation of quantitative risk impact.Although Program Evaluation and Review Technique (PERT) is widely used in risk analysis in project schedule, some deficiencies still exist. On the one hand, there is assumption in PERT, which is the duration of each activity obeys beta distribution. But the deviation of each beta density function and three time estimation can affect the precision of PERT computation to a great degree. On the other hand, there is a problem that the completion probability from PERT is always less than designated completion probability, and this problem has not been well solved. On the basis of previous studies, according to limited probability three-time estimation method, this thesis builds the minimum fitting variance models of beta and gamma distribution at first; and then puts forward the idea that to solve the model by using Particle Swarm Optimization (PSO); at last an engineering project is used to confirm a conclusion that for different types of activities, to choose suitable distribution patterns and to use the limited probability three-time estimation method can improve the accuracy and efficiency of PERT. On the other hand, this thesis builds upon the problem that the completion probability from PERT is always less than designated completion probability; combines particle swarm optimization (PSO) with PERT simulation method, and at last builds a feedback system which can change resource conditions of some activities to meet the requirements of completion probability. At last an engineering project is used to confirm that the feedback system is effective and practicable.