中文摘要：

通过估计失效事件的最优抽样概率密度函数,交叉熵法可有效降低失效事件的稀疏性,并显著提高电网可靠性蒙特卡罗仿真的收敛性。但目前交叉熵法局限于服从伯努利分布的随机变量,例如两状态线路失效模型,对于多状态离散随机变量（考虑降额的发电机）以及连续随机变量（风速、负荷等）尚待拓展。该文提出一种可以计及多状态离散型和连续型随机变量的扩展交叉熵法,给出了多状态离散型随机变量和连续型随机变量概率分布的交叉熵参数寻优方法,同时在此基础上融合预抽样阶段和最优抽样阶段的样本信息得到综合可靠性指标,进一步提升了电网可靠性评估扩展交叉熵法的计算效率。以IEEE-RTS79可靠性测试系统为例,考虑发电机降额状态和负荷随机波动,对含有风电场的电力系统进行可靠性评估,验证了该文方法的高效性。

英文摘要：

By estimating the optimal sampling probability density function （OS-PDF） for failure events, the cross-entropy method （CEM） has been utilized to overcome the rareness of failure events and accelerate the convergence of the non-sequential Monte Carlo simulation for power system reliability evaluation. The existing CEM can be applied to two-state discrete component state variables which follows Bernoulli distribution （such as the two-state state variable of transmission line）, but cannot be applied to multi-state discrete （such as the multi-state generator state variable including derated state） or continuous component state variables （such as wind state variables and load state variables）.Thus, this paper proposed an extended cross entropy method （ECEM）which was able to estimate the OS-PDFs for both discrete and continuous component state variables, to expand the application of CEM. Moreover, the proposed method used the system states sampled at both pre-sampling and optimal-sampling processes to calculate reliability indices, so as to enhance the evaluation efficiency. Considering the derated state of generators and the random variation of wind and load, the proposed method is tested by using IEEE-RTS79.