| The research of the seabed sediment types is always an important subject in researching area such as seabed mineral resources,marine geology,underwater engineering,marine fishery,underwater communications and so on.In order to provide better services for these areas,to improve further efficiency for marine development and resources utilization,it comes to be very important to develop the technology of seabed sediment recognition.Traditional sampling methods almostly have low efficiency,high cost and heavy workload.On the contrary the method of acoustic classification was improved,and it provides higher recognition efficiency,reduce the detection cost,and reduce the workload.The main objective of this paper is to study and practice the latest developed technology useing multi beam scattered wave data on seabed sediment classification.Multi beam measurement data mainly include depth data and backscatter intensity data.In this paper,taking multi-beam system EM122 measured original data as the research object,researching multi-beam's principle,selection of sediment acoustic recognition parameter,and solution data error sources were discussed and step-by-step analysis of the depth data processing reverse flow and backscatter intensity data processing flow.Then according to the characteristic of the final result data,this paper select the corresponding neural network to realize the recognition of the type.Main contents include introduceing the basic theory of acoustics,researching analytical processing of muilt-beam data and backscatter intensity data,expounding the necessity of using neural network in seabed sediment type recognition field,and study on using neural network to analysis backscatter intensity data.At the last,the SVM(Support Vector Machine)has been chose to practice neural network method,and verified the validity of this method of Sediment type recognition through instance data from the middle Okinawa trough. |
PDF Full Text Request