基于拟蒙特卡罗方法的电力系统可靠性评估  

A Quasi-Monte Carlo Method Based Power System Reliability Evaluation

作　　者：侯雨伸[1] 王秀丽[1] 刘杰 郭静丽[1] 唐伦[1] 

机构地区：[1]西安交通大学电气工程学院,陕西省西安市710049 [2]国网智能电网研究院,北京市昌平区102200

出　　处：《电网技术》2015年第3期744-750,共7页Power System Technology

基　　金：国家863高技术基金项目(2012AA052701);国家自然科学基金项目(51277140)~~

摘　　要：针对电力系统可靠性评估中蒙特卡罗方法误差收敛相对较慢的特点,将以低偏差点列抽样的拟蒙特卡罗方法应用于可靠性评估中。介绍了拟蒙特卡罗方法的原理,并在误差分析方面与蒙特卡罗方法进行比较。进一步介绍了Sobol低偏差点列的构造方法。建立了以Sobol点列为抽样点的拟蒙特卡罗模拟可靠性计算模型,并与重要抽样法相结合。对3机系统和RTS79测试系统分别进行标准非序贯蒙特卡罗模拟和Sobol点列拟蒙特卡罗模拟,计算指标EDNS与LOLP,并比较误差收敛曲线。结果表明,拟蒙特卡罗方法进行可靠性计算,EDNS和LOLP指标具有更快的误差收敛速度。该方法可作为电力系统可靠性计算的一种新思路。In allusion to slow error convergence of Monte Carlo method used in power system reliability evaluation, the quasi-Monte Carlo method sampled by low discrepancy sequence is applied in reliability evaluation. The principle Qf quasi-Monte Carlo method is presented and compared with Monte Carlo method in error analysis. Further, the construction of Sobol low discrepancy sequence is described. A quasi-Monte Carlo reliability calculation model, in which the Sobol sequence are taken as the sampling points, is established and combined with importance sampling method. For the three-machine system and RTS-79 the standard non-sequential Monte Carlo simulation and Sobol sequence quasi-Monte Carlo simulation are performed respectively to calculate reliability indices EDNS and LOLP, and the error convergence curves are compared. Comparison results show that when quasi-Monte Carlo method is applied in reliability calculation, the error convergence of indices EDNS and LOLP is faster. The proposed method can be regarded as a new thinking for power system reliability calculation.

关 键 词：可靠性评估 蒙特卡罗 拟蒙特卡罗 低偏差点列 

分 类 号：TM71[电气工程—电力系统及自动化]