Research On Data Fusion Technology In The Application Of Nondestructive Testing

Recently , with the development of computer software and hardware technology, data fusion technology has became a novel method in the signal processing. nondestructive testing technology have stricter requirements of data acquisition systems' speed, reliability and real time in order to satisfy the instant need of modern industry . Magnetic flux leakage testing method is widely applied for high Signal-to-Noise,high sensitivity and detection efficiency in nondestructive testing.Magnetic flux leakage technology is a nondestructive testing application based in electromagnetism, It has been used in steel and oil and chemical industry areas. It is the principle that magnetic induction lines in defect become deformation and bending and leak from surface defects after Iron magnetic material is magnetized .we detect Leakage magnetic field using sensor array , so it can make out something about defect. With the development of computer technology, process of magnetic flux leakage signal and intelligent recognition of defects are two important parts in pipeline magnetic flux leak detecting, The paper presents the significance of pipeline magnetic technology developing .For example, we must use multi channel sensor array that rounded highly with probe to detect flaw signal fully in large diameter and long pipe detecting.But the noise signal is mixed into detecting signal in the detecting process. We need to deal with detected signal by reducing noise signal to avoid inaccurate or missed detecting. This paper introduces an algorithm of signal de-noising by using wavelet Analysis. The point mutations of magnetic flux leakage signal contain the rich flaw information, so it is a good method to remove noise and improve Signal-to-noise ratio for using Wavelet Analysis.The pretreatment defect signal is taken to data fusion center for data fusion operation. This paper introduces that the flaw signal is detected by data fusion technology of radial basis function (RBF) neural network. While magnetic flux leakage signal is processed with RBF neural network, it mainly is to confirm network hidden layer nodes, data center of RBF function and adjust weights matrix from hidden layer space to output space. RBF network learning algorithm can be divided to two parts: identification of function centre and weight adjustment. After network hidden layer nodes, data center and standard deviation are generally confirmed, so network will be determined by learning weights. Artificial neural networks have learning, memory, lenovo thinking, parallel processing and other outstanding features, so it is widely used for Apparatus for calibration testing measurement and automation fault diagnosis.In multi-sensor measurement system, it can take many benefits for using data fusion technology, for example, enhancing System Stability, increasing system reliability and improving the detection capability. Compared with BP neural network, RBF neural network has better recognition performance, more accurate and effective in qualitative and quantitative analysis of the defect signals. The Simulation results show that RBF neural network improves identification of the MFL results and the detection ability and the precision of the signal.