Physical Interpretation of River Stage Forecasting Using Soft Computing and Optimization Algorithms

Abstract

This study develops river stage forecasting models combining Support Vector Regression (SVR) and optimization algorithms. The SVR is applied for forecasting river stage, and the optimization algorithms, including Grid Search (GS), Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and Artificial Bee Colony (ABC), are applied for searching the optimal parameters of the SVR. For assessing the applicability of models combining SVR and optimization algorithms, the model performance is compared with ANN and ANFIS models. In terms of model efficiency, SVR-GS, SVR-GA, SVR-PSO and SVR-ABC models yield better results than ANN and ANFIS models. SVR-PSO and SVR-ABC models produce relatively better efficiency than SVR-GS and SVR-GA models. SVR-PSO and SVR-ABC yield the best performance in terms of model efficiency. Results indicate that river stage forecasting models combining SVR and optimization algorithms can be used as an effective tool for forecasting river stage accurately.