Intraday Net Load Reserve Demand Assessment Based on CatBoost and Kernel Density Estimation

In the process of building a new power system with renewable energy as the mainstay, the uncertainty on both sides of the source and load is enhanced, and the power balance regulation faces serious challenges, which seriously threatens the safe and stable operation of the power system. The net load reserve demand assessment can help reduce the negative impact of new energy power fluctuation and prediction error on grid operation. In order to ensure the reserve adequacy of high percentage green energy power systems, this paper proposes a net load reserve demand interval prediction method based on CatBoost and nonparametric kernel density estimation method. First, the actual load/new energy power data and intra-day forecasts and other data are constructed into a multivariate time series to predict the net load reserve demand based on the CatBoost model. Then, a nonparametric kernel density estimation method is used to estimate the confidence interval of the net load reserve demand forecast error, and the interval of net load reserve demand forecast with a given confidence interval is derived. A test is carried out based on the data of the power grid in Shaanxi Province to verify the effectiveness of the proposed method.