Input optimization of ANFIS typhoon inundation forecast models using a Multi-Objective Genetic Algorithm

摘要: Abstract Inundation forecast models based on a subset of data inputs are advantageous in their efficiency due to the rather fewer needed be processed. This type is nevertheless rare mainly because difficulty selecting appropriate combination input variables. In this study, an innovative methodology proposed overcome by integrating Adaptive Network-based Fuzzy Inference System (ANFIS) and Multi-Objective Genetic Algorithm (MOGA). The three indices coefficient (CE), relative peak error (RPE) time shift (RTS) used assess performances from various perspectives. Optimal combinations for ANFIS searched MOGA, with best each index selected. Comparisons show that optimal obtained have better overall than ARX Nonlinear models. Test results reveal preferably selected designated prediction lead cannot maintain optimality under different leads. problem has been resolved use series models, optimized respect using same methodology. model exhibits significant improvements leads thus effectively improves accuracy inundation level prediction.