Font Size: a A A

Fast Reliability Evaluation For Power System Based On The Combination Of Important Sampling And Extreme Learning Machine

Posted on:2018-11-16Degree:MasterType:Thesis
Country:ChinaCandidate:P C XuFull Text:PDF
GTID:2322330518960944Subject:Engineering
Abstract/Summary:PDF Full Text Request
In view of the online assessment and control for power system reliability,it is necessary to evaluate the possible risk existing in power system based upon the real-time data.Due to quite a few uncertain factors as well as the grid size increasing,traditional Monte Carlo simulation(MCS)is no longer able to meet the requirements of real-time and high efficiency in the reliability evaluation of complex power system.Accordingly,combining with the characteristics of power system reliability assessment process,this paper analyzes the distribution features of the evaluate variables in power system reliability assessment,gives improvement measures to enhance the computing efficiency,and explores how to improve the computing efficiency of original MCS algorithm in different stages.First of all,this paper reduces the variance of LOLP by introducing an important sampling based on Cross-Entropy(CE)in the sampling part.The concept of dynamic fault set based on multiple index linked list is proposed,then combine it with important sampling for fast reliability evaluation of complex power system to accelerate convergence velocity of iteration.What's more,this paper presents a new algorithm which combines the Cross-Entropy based important sampling method with Extreme Learning Machine(ELM).On the one hand,the CE method is presented in state sampling to obtain the optimal probability parameters,leading to the decrease of the variance and speeding up the index convergence.On the other hand,ELM is utilized to state samples(including the component states and load loss)from important sampling for supervised learning.Then the network learning model,replacing conventional nonlinear programming,is used in state evaluation so as to realize rapid reliability evaluation.The proposed method is applied to IEEE-RTS 79,thus the calculation accuracy and efficiency of which is compared with that of crude MCS and the CE based important sampling method.As it is shown in the results,within a certain range of error,the proposed algorithm has better computation efficiency than crude MCS and the CE method significantly.
Keywords/Search Tags:Reliability evaluation, important sampling, Cross-Entropy, Dynamic Fault Set, Extreme Learning Machine, supervised learning
PDF Full Text Request
Related items