Primary Design And Research Of Fishery Detector

This dissertation roots in a part of 'Fish Exploring Instrument' scientific research fund item in Shanxi province government. To study signal processing techonology based on vector sensor is attracting more and more attention in underwater acoustic domain. The research content in this dissertation includes designing the multichannel data sampling system based on vector sensor and programming the software in order to recognise underwater target passively.The multichannel data sampling system based on vector sensor is composed of 8 data sampling channels to receive two vectors output to satisfy the demand of 'Fish Exploring Instrument' scientific research item; the underwater target passive recognition software extracts the feature of underwater target with two methods, underwater target classifier works with ANN (Artificial Neural Network).The multichannel data sampling system includes three parts i.e analog signal pretreatment circuit, data sampling circuit, data sampling control circuit. The band of Band Pass Filter, the multichannel plus in the analog signal pretreatment circuit can be adjusted; the data sampling control circuit controls the sampling frequency, the sampled data output timing ;the data sampling control program transfers the sampled data into computer and saves the data in order to go on data postposition processing.The underwater target passive recognition software is composed of the target feature extraction and the target classifier. The target feature extracting adopts Wavelet Packet Decomposing and Bispectrum Estimation these two methods. Wavelet Packet Decomposing extracts the energy features in different frequency band; Bispectrum Estimation processes the non-Gaussian characteristic in underwater signal, thus extracts the non-Gaussian features of underwater target. Designing the BP(Back Propagation) ANN realizes the target classifier.The multichannel data sampling system designed in this dissertation passedSonhua Lake in Jilin province and Fengjiashan reservoir in Shanxi province two experiments examination, the result indicates the system performance is stable and achieves the designed requirements. The underwater target passive recognition software processed the data acquired from experiment and get high ratio of underwater target recognition correspondingly, it showed the chosen methods to extract feature and the designed target classifier are efficient.