Short-Term Load Forecasting Model Using Flower Pollination Algorithm

Electricity is natural but not a storable resource and has a vital role in modern life. Balancing between consumption and production of the electricity is highly important for power plants and production facilities. Researches show that electricity load consumption characteristic is highly related to exogenous factors such as weather condition, day type (weekdays, weekends and holidays etc.), seasonal effects, economic and politic changes (crisis, elections etc.). In this study, we propose a short-term load forecasting models using artificial intelligence based optimization technique. Proposed 5 different empirical models were optimized using flower pollination algorithm (FPA). Training and testing phase of the proposed models held with historical load and weather temperature dataset for the years between 2011-2014. Forecasting accuracy of the models was measured with Mean Absolute Percentage Error (MAPE) and monthly minimum approximately %1,79 for February 2013. Results showed that proposed load forecasting model is very competent for short-term load forecasting.