Using clustering to evaluate reliability of composite power systems

A methodology is proposed to incorporate wind farms into a conventional power system which includes transmission constraints at composite system level. Monte Carlo method is used to represent operating states of turbines over time. The methodology uses clustering to consider correlation between load and wind power and sensitivity analysis to calculate reliability indices of the system efficiently. We validate our approach by using the IEEE Reliability Test Systems as well as the wind speed data from the National Renewable Energy Laboratory. It is shown that clustering can give more accurate results than the exact transition rate method previously proposed.