A systematic second-order output spectrum based method for fault diagnosis with a local tuning approach

In order to diagnose potential serious and initial damages (loose bolts or fatigue cracks) in industrial structures more precisely, a systematic second-order output spectrum (SOOS) based method with a local tuning approach (LTA) is proposed in this article. In the novel method, structures with faults, material nonlinearities and nonlinear boundary conditions are represented by the Volterra series, and then a general ring type multi-degree-of-freedom (MDOF) model simulating nonlinearities as nonlinear restoring forces is built to describe the structures’ behavior. With defined nonlinear fault features, two local damage indicators based on the SOOS and LTA are derived for diagnosing serious and initial faults in sequence. Results of experiments on a satellite-like structure with loose bolts show that the systematic SOOS based method can identify serious and initial bolt loosening faults more accurately. Thus, it can be more applicable for fault diagnosis in complex structures in practice.