Analysis of acoustic emission signals in machining

Acoustic emission (AE) signals are emerging as promising means for monitori ng machining processes, but understanding their generation is presently a t opic of active research; hence techniques to analyze them are nor completel y developed. In this paper we present a novel methodology based on chaos th eory, wavelets and neural networks, Sor analyzing AE signals. Our methodolo gy involves a thorough signal characterization, followed by signal represen tation using wavelet packets, and state estimation using multilayer neural networks.; Our methodology has yielded a compact signal representation, fac ilitating the extraction of a tight set of features for flank wear estimati on.