Abstract—In power system studies, load duration curve (LDC) plays an important role in medium term horizon power system planning, reliability studies, energy markets, and economic analysis of electric power systems. Therefore, forecasting seasonal LDCs is beneficial to network operators as well as the market participants. Finding a simple and accurate model to forecast LDC is an important issue. This paper proposes a new framework for forecasting seasonal LDC. As there are few contributions regarding forecasting curve time series, we redefine the problem of forecasting LDCs into vector forecasting problems. In fact, we divide LDCs into three parts, and then, artificial neural network (ANN) engines are used to forecast future values of these three parts. The load data of Alberta electricity market from 2000 to 2013 is used to verify the validity of the proposed method.

Keywords