Short-Term Electric Load Forecasting at Southern California Edison

摘要

This paper discusses Southern California Edison's (SCE) experience with electric load forecasting for energy trading, planning, and scheduling purposes. We first describe the various forecast models being used and the forecasting processes conducted by the SCE short-term demand forecasting team. We then focus on day-ahead load forecasting and introduce our inhouse statistical models. The forecasts generated from the inhouse models are compared with forecasts from two commercial load forecasting software packages. It is shown that the in-house methodologies outperform the two vendor models.