Hybrid Genetic Algorithm and Local Search for Energy Demand Prediction Model

Energy demand pattern have many variables related to uncertainty behavior. These lead to a higher estimation rate of energy demand forecasting. However, two problems need to be overcome. The first problem is the fitness evaluation in energy demand forecasting model in which more than one variable are included, and the second problem is the local optimality that single algorithm fails to solve. The objective of this research is to develop energy demand forecasting model that reflects the characteristics of energy demand. A local search is used to assist the genetic algorithm in overcoming uncertainty in demand and the local optima problem and thus producing a higher estimation rate. To evaluate the performance of energy demand model, the actual demand was compared to estimation results. The findings indicate that the solution obtained using the proposed model was an improvement in quality over that obtained by a single genetic algorithm and can be applied to forecast future energy demand with higher approximation accuracy.