Modeling Conditional Forecast Error for Wind Power in Generation Scheduling

The integration of wind power requires additional operating reserves to cope with the uncertainty in power system operation. Previous research shows that the uncertainty of the wind power forecast varies with the level of its output. Therefore, allocating reserves dynamically according to the specific distribution of the wind power forecast would benefit system scheduling. This paper presents a statistical model to formulate the conditional distribution of forecast error for multiple wind farms using copula theory. The proposed model is tested using a set of synchronous data of wind power and its day-ahead forecast. It is then utilized in a stochastic unit commitment model to simulate the day-ahead and real-time scheduling of the modified IEEE RTS-79 system integrating wind power. The results show that scheduling reserves dynamically according to the modeled conditional forecast error reduces the probability of reserve deficiency while maintaining the same level of operating costs.