Structure noise reduction of ultrasonic signals using artificial neural network adaptive filtering

A multi-layer linear neural network within the framework of adaptive FIR filtering is presented for enhancing the signal-to-noise ratios of ultrasonic signals. The backpropagation (BP) algorithm is used to train and adjust the network weights. From the results obtained on large grained materials and on composites, it is apparent that the performance of the new approach is better than those of conventional adaptive algorithms in reducing the structure noise of ultrasonic signals. (C) 1997 Elsevier Science B.V.