Short term load forecasting in power systems using a hybrid approach based on SVR technique

Optimal operation management of distribution systems requires suitable tools for accurate forecasting. By the appearance of new technologies such as electric vehicles, renewable sources, etc, the variation of the electrical loads is intensified. In order to solve this issue, a new powerful forecast method based on support vector regression (SVR) is proposed to model the behavior of load consumption. According to the high complexity and severity of the problem, a new optimization approach based on firefly algorithm (FA) is proposed to adjust the optimal parameters of the SVR. The new version of FA called θ-FA is used to search the solutions in the polar space instead of the traditional Cartesian space. The objective function is to minimize the average forecast error considering the SVR constraints. The feasibility and high performance of the proposed method are examined on practical test data.