Load forecasting assessment using SARIMA model and fuzzy inductive reasoning

Abstract

Accurate load forecasting is critical for power systems planning, control, and operation. Poor forecasting in volatile power markets can have large, detrimental impacts on power system costs and real-time energy acquisition costs from distribution companies. This paper implements and compares two different methodologies for short term load forecasting: a classic statistical model (SARIMA model) and a model based on artificial intelligence (Fuzzy Inductive Reasoning, or FIR, model). A numerical example predicts one week for every methodology and the results are compared for both models.