Study On Variance Reduction Technique For Power System Reliability Assessment

This dissertation is supported by National Natural Science Foundation of China (No.50607021).Monte-Carlo simulation is more available and flexible in the power system reliability assessment. Compare to analytic method, it can simulate the real operating conditions of power system, and find out a lot of operating and failure states and is good at the simulation of faults that are composed of multiple,related and linkage failure states. Because these features of Monte-Carlo simulation, it is good at the composite system reliability assessment and is becoming an important theoretical method that is more of a concern.But there is a key contradiction that is the more accurate, the more time. If a high accuracy of reliability indices is demanded, large-scale sampling calculations need to be carried out and a great of time must be spended. This shortcoming limits the application field of this method.Through reducing the variance of sampling samples, the samples are decreased and the convergence speed of the simulation is accelerated at the same accuracy. The variance reduction techniques which are using are: importance sampling, control variates and antithectic variates etc. The importance sampling can reduce the samples'variance to zero in the theory, and is considered to the most effective method. But there is a deadly defect in the traditional importance sampling, it is that the different components'effects to the power system reliability are neglected, so the convergence speed of the simulation is not further improved.A sampling method named splitting optimal multiplier based on the importance sampling is proposed according to the defect. The components are sampled separately according to their effects to the power system reliability in the method. The convergence speed of the presented method is verified using the IEEE-RTS79 test system. The main points of the method are the following:â‘ The importance identification is realized using the product of a component's capacity and its forced outage rate based on the component's different effects to the power system reliability. Before the Monte-Carlo simulation, the importance identification has been finished.â‘¡The simulation optimization methods are introduced and used in the solution of the optimal multipliers that is different from the traditional importance sampling. It is that the optimal multipliers for importance components have been searched through golden section searching approach while the Monte-Carlo simulation is proceeding. Then the importance sampling is carreid out among these importance components while the traditional Monte-Carlo simulation is done among the remaining components.â‘¢The aim of this paper is not only the reduction of the samples'number but also the decrease of the simulation time at the same accuracy. So the convergence speed of the Monte-Carlo simulation is further accelerated.â‘£The layer optimization is presented:the optimal multipliers of components in power system are searched in a sequence according to the components'important extent to the power system. So the issue of multivariate distribution's optimizaion is transferred to the issue of many steps'optimization,the convergence speed of the Monte-Carlo simulation is accelerated.