Abstract:

This paper applies wavelet analysis and neural networks to composite materials non destructive test (NDT). Using the good time frequency characteristic of wavelet analysis, characteristic information can be picked up from high noise. An improved ART neural networks algorithm is presented. Experimental results show that applying wavelet analysis and neural networks to the damage test of composite structure can effectively detect the smashing position and intention when smashing appears.

Key words: wa velet analy sis, neural netw or ks, composit e material, smashing