Study On Signal De-noising Technology In Ultrasonic Testing Based On Independent Component Analysis Theory

In NDT(Non-destructive Testing) of weld flaw, UT(Ultrasonie Testing) is an important method. In Ultrasonic NDT, the background noise by the unresolved scatterers such as grain boundaries and other mierostructure, and electrical noise by the ultrasonic defect detector often masks the flaw signal, creating a hindrance to deteetion. All of these can affect the ultrasonic testing result. Independent component analysis (ICA) is a new blind separation of sources testing technique. It deal with a group of mixed signal cobinated by some specialty source signals and purpose to separate these specialty source signals frome mixed signal.Through this way we can get rid of noise signals. The primary objective of this paper was to investigate the effect of ICA used in separating inspect signal and source signal.Firstly, the mathematical model and principle of ICA are studied, and the different independent criteria and several main algorithms of ICA are discussed. Further, the Fast ICA algorithm is studied. Fast ICA is a fast algorithm of ICA, which is based on Fixed-point iteration theory to fix the non-Gaussian maximum. Fast ICA algorithm parallelly processes a large amount of sample point of received signals via Newton iterative algorithm, and recovers one independent component from the receiving signals one time.Then some simulation experiments were carried out to find the difference between FastICA algorithm,JADE algorithm and AMUSE algorithm in signal separating , and the results showed that the FastICA algorithm had good properties in both Convergence Rate and similarity coefficient. So it can be used in signal de-noising of automatic ultrasonic test in practical T-shape laser welding joint. Based At last, separating noise from observed signals was studied in this paper when the T-shape laser welding joint was inspected by ultrasonic testing system adopting independent component analysis theory to process the signals. The principle of automatic ultrasonic testing signals processing and negentropy law of ICA were introduced. The experimental data were processed using relative analysis tools. We picked the mixed figure signals up from the scan wave and separated the source signals and noise signals, after separation, we got a echo signal and noise signal clearly. Results showed that the FastICA could separate defects signals from noise effectively in laboratory.