A hybrid intelligent approach for modeling nonlinear complex data: A case study, prediction, Southwest Iran

One of the significant issues in the area of engineering is the modeling complex data with the target of prediction, classification, clustering, etc. Nevertheless, by the rapid growth of the technology, the high complexity and nonlinearity of the existing data is increased such that they put most of the traditional forecast models such as linear and nonlinear models prone to bias. In order to solve this issue, this paper suggest a newly introduced forecasting model called support vector regression (SVR) model to reach more accurate modeling and thus prediction of the complex data. In order to adjust the setting parameters of this model, the clonal selection algorithm (CSA) as a powerful optimization tool is employed. In addition, a sufficient modification is proposed to improve the search ability of this algorithm effectively. The practical test data (from 230 kV substation of Southwest Iran) are employed to demonstrate the satisfying performance of the proposed hybrid model.