Signature Analysis for Reciprocating Machinery with Adaptive Signal Processing

The spectrum analysis by means of FFT (fast Fourier transform) analysis has been widely used in mechanical signature analysis. An alternative method is proposed in the paper for condition monitoring and diagnostics of reciprocating machines. A modified LMS (least mean square) algorithm is proposed in order to model the impulse-like signals with non-stationarity. Convergence of the algorithm is investigated. The approach presented is suitable to real-time monitoring during the process of working operation and fault diagnostics of reciprocating machines.