It is of great importance to make a proper classification of underwater sediments for protecting the ecological environment and dredging materials such as sand or gravel, mud, etc. in rivers or oceans.
Ri Un Song, a section head at the Faculty of Shipbuilding and Ocean Engineering, has newly defined some characteristic parameters and established Artificial Neural Network (ANN) with these parameters as input layers. By doing so, he has managed to find a way to raise the accuracy of acoustic sediment classification on rough riverbeds.
The neurons of input layers used include the newly-defined roughness and monotonic decreasingness of the tail potion of the first echo signal as well as the roughness index and hardness index usually employed in Rox Ann and QTCView.
The proposed method has raised the accuracy of acoustic classification of sediments in the water areas with rough floors to 95%.
It is now in effective use for search for gravels and sand by dredging vessels on riverbeds, and for classification of sediments accumulated on the dams of floodgates, tidelands, etc.
© 2021 KumChaek University of Technology