A Study of Autoregressive Conditional Heteroscedasticity Model in Load Forecasting

Abstract

In this paper, ARCH (autoregressive conditional heteroscedasticity) effects of load time series is analyzed and a new feasible method of load forecast based on ARCH model is proposed. The main achievements of the paper involve the following aspects: Firstly, by testing the disturbance series of the classical model, Some witness to the ARCH effect is reported by LM test. Secondly, An ARMA-ARCH model is constructed. This model is judged to have good forecast ability, and compares favourably with ARMA model in theoretic significance. Furthermore, a relatively deep research on the second-order moment of load time series proceeds from such a viewpoint as reverse leverage effect, the mechanism of asymmetric effects between different shocks is demonstrated. Finally, we enhance the ARCH model to a new level with the help of extended ARCH class model. These models have more practical preconditions, and have more powerful applied significance than some classical models. It may be a promising new method for short-time load forecasting.