| Rotary machinery is widely used in mechanical engineering. It plays an increasingly important role in various industry fields. Damages caused by some linear or non-linear factors pose great challenges to the safe operation of rotary machinery. Thus it is quite significant to carry out damage detections on rotary machinery. It is easy to get to know the occurrence of damages while the real difficulties lie in how to locate and quantify damages.A rotor is the core component of a rotary machine. In this research, in order to localize and quantify damages of a rotor, the loss of bending moment is reckoned as the damage index. Piezoelectric Impedance Method(PIM) and Spectral Element Method(SEM) are introduced in the modeling. As is known, structural health monitoring based on PIM benefits from the fact that mechanical impedance changes in accordance with the changes in the structural integrity. Thus the changes in structural integrity can be revealed by the changes in mechanical impedance. Moreover, the spectral element method combines the highly-computational efficiency feature of spectral method and finite element method. In this paper the focus of the research is placed in the numerical model building based on PIM and SEM. For the special circumference boundary condition of a rotor, annular elements are introduced as the analyzed elements. Based on the E/M coupled impedance analysis model, numerical expressions of damaged and undamaged rotors are obtained by using related parameters and deduced formulas. Lastly, a Taylor-expansion method is used to obtain the approximated damage index and identification algorithm is formulated. To validate the proposed method, a numerical case study and an experiment are carried out. Both the location and the severity can be obtained and the accuracy and efficiency is thus proved. For the impedance test experiment, de-noising problems are also considered. Thereby, a second-order adaptive stochastic resonance method is proposed and used for the de- noising. A second-order model is used and an iterative computation method is proposed to search for the optimal system parameters. Simulation and experiment results have shown that good filtering performance is achieved and larger SNR is obtained with higher –computation efficiency. |
