Research And Implementation Of AUV Underwater Intelligent Detection System

In recent years,autonomous underwater vehicles(AUV)have made rapid breakthroughs thanks to the development of battery and intelligent technology.Various countries have vigorously developed this emerging water vehicle,which has rapidly increased the number of AUV and seriously threatened China’s offshore security.Therefore,early underwater detection and early warning of AUV are becoming more and more important.When the AUV is operating,the propulsion motor and propeller will radiate their unique sound signals into the surrounding water.Passive underwater acoustic detection relies on hydrophones to convert underwater acoustic signals into electrical signals for classification and identification,which is an important method for underwater target recognition.Therefore,this subject mainly develops a low-cost passive underwater acoustic detection system for AUV,and mainly completes the AUV automatic warning and target classification functions in shallow waters.This article first summarizes each important workflow of the underwater acoustic detection system,and on this basis,determines the overall composition of the system.The propagation model of sound in water is analyzed,and several common hydrophones are compared according to the characteristics of underwater acoustic transmission.The sensitivity curves of hydrophones are measured experimentally to complete the selection of hydrophones.Secondly,the underwater background noise and motor propeller noise were actually collected and visualized,and the characteristics of different noises on the spectrum were analyzed.Based on this,the adaptive spectral line enhancement(ALE)algorithm was used to control the background noise signal.By improving the algorithm,the convergence speed is increased,the steady-state error is reduced,and the signal-to-noise ratio of the processed signal is significantly improved.Third,in terms of feature extraction,the wavelet packet transform algorithm is improved,and the energy features of the signal are extracted as the basis for classification.At the same time,the time-frequency feature extraction algorithm of the short-time Fourier transform is optimized,and the former is used as the main feature extraction algorithm after effect comparison.Fourth,completed the signal acquisition system based on the XMC4500 processor,transplanted the FreeRtos system,completed the underlying drivers such as SD cards,ADC,and the underlying software of the multithreaded acquisition card,enabling it to collect underwater acoustic signals of different durations with multiple sampling rates.Fifth,different classification algorithms such as k-nearest neighbor and SVM are introduced,and a hybrid classifier suitable for underwater target detection is designed according to their different characteristics.The experimental results show that the algorithm has a better recognition effect.Finally,the shortcomings of this system are summarized and the future exploration is prospected.