Hybridizing the Machine Learning Techniques for Prediction of Sediment Yield

Abstract

Rivers are an integral part of the hydrologic cycle and are the major dynamic geologic agents that play major role for transformation of sediments from land to the oceans. Sedimentation is the biggest problem. Evaluation of suspended sediment yield is an essential parameter under the assessment on Dam filling, protecting of aquatic organism and wildlife habitats, understanding the flood capacity and hydroelectric equipment in hydro-electric power. The assurance of sediment yield through different traditional way isn’t exactly correct because of the participation of different complex processes.There are many limitations of traditional methods but it can be overcome by artificial intelligence techniques. So, in this study, the MOGA-ANN (Multi-objective genetic algorithm based artificial neural network) hybrid artificial intelligence method is used to estimate the sediment yield in Krishna river basin, India. The research done for evaluation of the suspended sediment load by taking 20 years of data from Vijayawada, gauging station which is the downward station in Krishna river. The proposed MOGA-ANN model provided low root mean square error (0.03354) and high correlation coefficient (0.9214) during test phase. It exhibited satisfactory performance.