Stochastic economic dispatch strategy based on quantile regression

Abstract

This paper proposes a stochastic economic dispatch strategy considering uncertainties of source and load based on difference regional energy resources. A improved autoregressive model is used to fit output and forecast error data of wind, photovoltaic, and load output for reproducing the asymptotic distribution and diurnal variation. The improved model generates a large number of deterministic scenarios to display the uncertainty and intermittency. Considering uncertainties of source and load based on difference regional energy resources, the quantile scenario reduction method is proposed to obtain different scenarios for ensuring the stochasticity of renewable energy and load. The quantile regression principle in statistics is used to set weights for different quantile scenarios. The proposed multi-scenario stochastic dispatching problem is converted into a MILP model for solving. Some conventional generators of the standard IEEE 39-bus system are connected with renewable power sources for study purpose. Actual data of a regional grid of China is analyzed. The simulation results of the model and strategy are summarized, analyzed and compared in this study. The simulation results show that the cost is reduced by 1.78% and 1.692% compared with the deterministic case and the traditional method. The proposed method could meet its different dispatch requirements and provide a theoretical basis for the generation of dispatching plans.