Comparative Study on modeling Efficiency Between Support Vector Machines (SVMs) model and Parallel OBF-NN model

This project is about the comparative study between model efficiency between support vector machine (SVM) and parallel OBF-NN model. To demonstrate the concept, basic support vector regression (SVR) model is developed as nonlinear model identification. Best parameter and option for SVR model is selected in order to construct optimum model performance. The study is developed using selected case study, which is using van de vusse reactor datasets. The data consist of input and output than applicable to perform simulation as training and validation data. Lastly, an OBF-SVR model is developed that use OBF model as linear part and SVR model as nonlinear part align in parallel. The performance of each developed model is tested in their performance in validation to approach real system value. The developed OBF-SVR model is compared with OBF-NN model and the deviation between each model is investigated.