Maintenance costs constitute a major part of operating costs in any industry. This has motivated industries to adopt best machine condition monitoring techniques so that costs can be reduced and productivity can be increased. In many rotating machinery system, shaft cracks are frequent and serious malfunctions that may lead to catastrophic failure and financial loss. Fault detection techniques are crucial for safe and reliable machinery and the demand for timely and accurate detection is increasing day by day. Shaft crack detection is an especially critical task due to the complex operating conditions that shafts are subjected to. One of the shaft crack detection approaches is vibration and acoustics monitoring, which can be implemented in an automated fashion. This paper presents the application of recurrence plots (RPs) and recurrence quantification analysis (RQA) in the diagnostics of various rotating shaft cracks. The RP is a two-dimensional visualization technique to investigate high-dimensional dynamical systems. It identifies the times when the state space trajectory of the dynamical system visits roughly the same area in the state space. On the other hand, RQA offers a more objective and quantitative method for the investigation of dynamical systems, which will represent the extracted features that characterize the system response. In addition, RPs and RQA are a modern tool for nonlinear data analysis, which enables us to investigate the various responses of the system (i.e., periodic, quasi periodic and chaotic) and provides valuable information about the dynamics of the system. The computed RQA features are ranked and the optimal set is selected using mutual information. Finally, an artificial neural network is used as a classifier to distinguish between the different shaft conditions.
A laboratory scale rotor test bed was used to investigate shaft crack detection techniques under controlled conditions. The study was implemented on a shaft that was seeded with two damage conditions produced by a crack propagator over 24-hour and 48-hour time periods. The horizontal and vertical displacements were measured for each shaft condition using proximity probes.
The study demonstrates that the RQA provides rich information about the status of the health of the shaft. Furthermore, results show an outstanding performance of the RQA in shaft crack detection with minimal knowledge about the dynamic response of the system.
