Research On Acoustic Source Localization In Enclosed Liquid Field

Large-scale vertical metal tank is one of the most important facilities of oilstorage. Corrosion of bottom palate is the most primary and hidden factors that affectthe security of oil storage tanks. It’s is meaningful to study on advanced tank bottomcorrosion online testing technology for maintaining stable operation of tanks. Thispaper presents a method of oil tank bottom corrosion testing called underwatercruising sensor system. Acoustic source localization was one of the key problems forthe underwater cruising sensor system. The problem was summarized in this paper.The problems of acoustic source localization in enclosed liquid field were studied inthis paper and the corresponding solving methods were proposed. The study includesthe following aspects:(1) Domestic and foreign situation of oil storage tank bottom corrosion detectionwas analyzed. Underwater cruising sensor system was proposed to measure the innerparameters online in enclosed liquid field. The positioning problems in enclosedliquid field were analyzed.(2) The nonlinear of positioning equations of TDOA (time difference of arrival)and bias of system model affect the localization performance of acoustic source inenclosed liquid field. The optimized EKF (extended Kalman filter) based localizationmethod and PF (particle filter) based localization method are proposed. For theoptimized EKF method, the adaptive fading factor was used to prevent algorithmdivergence. The EKF was optimized by adjusting the current measurements with theformer localization result. For the PF based algorithm, the state space model was usedto descript the localization problem. The posterior density of time delay estimationwas processed iteratively to eliminate or weaken the fake peaks caused by themultipath. The two algorithms have smaller deviation and better ability of stabilitycompare to the traditional methods and is a powerful means of acoustic sourcelocalization in enclosed liquid field.(3) An algorithm of generalized cross correlation based on extended AMDF(Average Magnitude Difference Function) was proposed to improve the accuracy oftime delay estimation (TDE). The effect of multipath was eliminated with thecontinuation of signals. The GCC was weighted with the AMDF and PHAT. Theexperimental results with different signals and different multipath showed that the algorithm can overcome the effect of multipath and noise in enclosed liquid field andimprove the stability and accuracy of TDE.(4) The time of flight (TOF) was used to locate the underwater cruising sensorsystem in enclosed liquid field. The algorithms based on GCC of spread spectrumsignal and principle component analysis (PCA) were proposed to improve theperformance of TOF estimation. In the former algorithm, the emitter signal wasmodulated by the direct sequence spread spectrum with m sequence. The receivedsignal was demodulated and the index of the peak of GCC was treated as theestimation of TOF. In the latter algorithm, the covariance matrix of signals wasdecomposed to get the signal subspace composed of signal feature vectors and noisesubspace composed of noise feature vectors. The accuracy of TOF estimation wasimproved by selecting the eigenvalue and eigenvectors which corresponds to the maininformation of signals and eliminating the redundancy item caused by the coherentmultipath and noise.(5) Simulation experiments were performed to validate the proposed algorithms.The underwater cruising sensor system and experimental system which can analysisthe waveform of acoustic source and perform the localization algorithms weredesigned and verified the performance of the proposed algorithms.