The intermittency and randomness

which are typical operational characteristics of wind power generation

have brought about negative effects to the reliability of power system operation. Monte Carlo simulation(MCS)based tools have proved useful in reliability evaluation. However

they face difficulties given the randomness of the fluctuant power output while the network is highly reliable particularly with the development of the power system. An improved cross entropy(CE)method is proposed which combines the CE and the dynamic fault set. The CE is presented to obtain the optimal probability parameters leading to a speed-up in state sampling efficiency. In addition

a form of multiple index based on link-list is proposed to replace time-consuming optimal power flow(OPF). Finally

the proposed method is well applied to composite power system reliability evaluation tested on a modified IEEE RTS-79 system. The efficiency of MCS

CE and the proposed method are compared in systems with different adequacies by altering the consumer load.