Study On Corrosion Acoustic Emission Separation Based On Blind Source Separation

When tank bottom is under corrosion detection, comparing with traditional detection methods: ultrasonic detection, eddy current detection, magnetic flux leakage detection, acoustic emission detection has these advantages: without empting the tank, low cost and it can achieve real-time detection, In one word, acoustic emission detection is an ideal tank bottom corrosion detection method, This paper includes: generation mechanism of corrosion signals,acoustic emission detection experiment,separation of mixed signals based on blind source separation and wavelet decomposition of corrosion signals.Corrosion acoustic emission signals can be divided into two types: burst and continuous. The paper analyzes generation mechanism of corrosion acoustic emission signals and influence factors such as metal surface passivation film rupture and chemical reaction process of metal ions. PIC-2 acoustic emission acquisition system is used to conduct experiment study and the experiment uses 10% FeCl3·6H2O solution as corrosion solution. During experiment process, waveforms of corrosion acoustic emission signals are recorded real-time in order to be processed in order to obtain exact information. Experiment results demonstrate that: events of acoustic emission signals show different trends with the corrosion continues, and with the increasement of corrosion time, the number of acoustic emission events also increase. The experiment result can be a reference for corrosion acoustic emission detection of actual tank bottom.Wavelet decomposition is used to analysis burst and continuous acoustic emission signals. Signals are decomposed by different wavelet scales. Results show that: energy of burst acoustic emission signals concentrates, mainly between 50kHz~80kHz. Burst acoustic emission signals have intermittent, the reason is energy release caused by metal surface passivation film rupture and bubble rupture; energy of continuous acoustic emission signals relatively disperse, mainly between 30kHz~50kHz, the reason is energy release by chemical reactions during corrosion process. This paper also studies separation of signals, Because blind source separation can achieve signals separation under the condition the number of source signals is unknown, ybrid iterative separation algorithm of non-Gaussian signals is designed. Separation software based on Matlab platform is also designed. At last, separation of mixed signals has been achieved.