Paper Title
Parameters Estimation of Linear, Quadratic and Exponential Models for Energy Demand Forecasting by Using ABC Algorithm

Abstract
The accurate energy demand forecasting is important issue for energy supplier companies. Especially, the electrical energy price is defined one day ago depending on energy demands. Developed accurate model increases management capacity and efficient usage of power plants. Despite the fact that, In Turkey, transformer based energy measurement for a specific region is still developing topic and application, the consumption data for a specific region can be very limited. In cases where there is limited data, modeling is a difficult task to perform. Although there are several methods have been proposed before, they suffers from forecasted values close to real ones efficiently. In this study, a model was developed to forecast electrical energy consumption in a specific industrial region in Turkey. In the model, the energy consumption of an industrial region between years of 2014 – 2016 was considered as output values and temperature and energy consumption values belonging one day ago are defined as input values. The output values were estimated by giving input values to the model, which is based on linear, quadratic and exponential functions. In the study, parameters of these functions were defined by Artificial Bee Colony Algorithm (ABC) by minimizing the error value between predicted and real values. Results obtained from three different functions were compared with each other and shown as graphically. Results show that the ABC algorithm can define the parameters of functions very accurately by decreasing the error value very close to zero. Index Terms - Energy demand forecasting, Artificial Bee Colony, parameter estimation, linear regression