【摘要】 为了解决蒙特卡洛法的模拟计算精度与计算速度的矛盾，研究了蒙特卡洛法的模拟计算误差与随机数序列周期的关系，提出了蒙特卡洛法全周期抽样的新概念，从理论上证明了全周期抽样的收敛解是惟一的，与随机数种子无关，与系统规模无关．通过对ＩＥＥＥ－ＲＴＳ可靠性试验系统的计算，说明全周期抽样蒙特卡洛法优于状态枚举法，其误差值随着随机数序列周期数的增大而减小．对不同规模的系统，选取周期数合适的随机数序列，全周期抽样所得到的计算结果就可以达到一定的精度，而不必采用周期数很大的随机数序列，从而提高了蒙特卡洛法的模拟效率．更多还原

【Abstract】 In order to improve the computation speed of Monte Carlo simulation, the relationship between random number generator and Monte Carlo simulation convergency is studied and the new concept of complete sampling for Monte Carlo simulation is presented. The outcome of Monte Carlo simulation founded on complete sampling is unique and independent of random number seeds and system dimensions, which has been proved theoretically. The computation to the IEEE-RTS shows that Monte Carlo complete sampling method is better than enumeration approach and the variance decreases with the increase of the period of the random numbers. Thus, to improve the Monte Carlo simulation convergency for different systems, we can choose appropriate random number generator under the given variance.更多还原