Investigation and Validation of Flow Characteristics Through Emergent Vegetation Patch Using Machine Learning Technique

Abstract

Investigation on change in flow characteristics in an open channel flow is important to understand the flowing water ecosystem, which is important in various aspects like sediment deposition and water quality. A significant number of field and laboratory experiments have been carried out to characterize different types of flows in an open channel. Machine learning technique learns automatically from the given experimental data set and also improves from given data without explicitly being programmed. For years, various laboratory and analytical experiments have been carried out to establish the turbulent flow characteristics change. To validate the laboratory experimental data of turbulent flow characteristics with machine learning techniques is scanty in the literature. Here in the laboratory experiment, a wide rectangular channel with an emergent rigid vegetation patch has been used, and the velocity of water has been measured in different locations and directions. Out of three sets of laboratory experimental data, two sets of data have been considered as training data, and by using “Polynomial Regression Techniques” we validated stream-wise velocity of the third set of data. Interestingly, it has been found that both experimental and theoretical data closely matched with each other. This new approach using the machine learning technique may become important for future research in all types of vegetated flow.