Analisis penentuan jenis kawanan ikan berdasarkan deteksi fasa pantulan gelombang akustik dan penerapan hidden markov model

Abstract

The aim of this research was to show that detection of phase shifted of acoustic wave reflection due to the movement of the fish school could identify the kind of fish school or the moving targets from its properties. Using the Hidden Markov Model (HMM), the observed fish school could be recognized from other fish schools. The research were executed into three experiments, namely phase shifted detection simulation, interference simulation and in situ experiment. The phase shifted detection simulation was executed to prove that the waveform of phase shifted reflected acoustic wave of moving target could identify its characteristics based on its dimension, shape, speed and structure in a group that simulated by the number of imitation fishes. During experiment, the imitation fishes were pulled by electric motor with different speed and transducer positions toward the direction of imitation fish. The research also analyzed the influences of several kind of possible interferences generated during the phase detection of the acoustic wave reflection. The simulation were done by means of electronic circuitry that represented .the fishes wave reflection and by the interference waves such as the reflection wave, and the engine sound. In situ experiment were executed to observe the real condition of the phase shifted reflection wave for 3 (three) kind of fishes. The three kind of fishes, consist of schooling fish species, shoaling fish species and a solitary fishes species . The phase shifted data were collected from the experiments in the fishing pool in the Seaworld and in a 4x4 m sea cage in Genteng Island. In this research, the phase shifted wave of all kind of fishes and all kind of imitation fishes was be detected by the fish species detector using a phase shifted detection technique. To distinguish the kind of fishes or imitation fishes, the angle between two or more line slope of the spectrum envelope of each phase shifted waves were calculated. If the two lines were parallel or the angle between the lines were 0o, the spectrum or the phase shifted wave were coincide each other, or the two wave were the same and, if the two lines formed a certain angle, then such spectrums or phase shifted waves were different. Hidden Markov Model (HMM) technique was applied for recognition processing. Computational results, showed that the reflection wave of the schooling fishes could recognized the fish school accurately (the fish school were 100 % identified successfully), however such recognition processing was not valid for shoaling and solitary fishes. From the interference simulation results, the reflection wave due to the surrounding areas of the objects, engine noise, fish direction and transducer position were not affected the result of identification and recognition processes.