Analysis of the single stage cycloidal gearbox with lobe defects. Fault diagnosis attempts using coherence function and Morris minimum-bandwidth wavelets

In this article, the frequency characteristics of the forces and torques in the various cycloidal gearbox designs were investigated. The aim of the article is the search for frequency patterns that could be used in the formulation of a fault diagnosis methodology. Numerical analysis was performed in the cycloidal gearbox without defects as well as in cycloidal gearboxes with lobe defects or with removed lobes. The results of the numerical analysis were obtained in the multibody dynamics model of the cycloidal gearbox, implemented in Fortran and using the 2nd-order Runge-Kutta method for the integration of the motion equations. The used model is planar and uses Hunt and Crossley’s nonlinear contact modelling algorithm, which was modified using the Heaviside function and backlash to fit cycloidal gearbox model convergence demands. In the analysis of fault diagnosis methods, the coherence function and Morris minimum-bandwidth wavelets were used. It is difficult to find a unique pattern in the results to use in the fault diagnosis because of the random characteristics of the torques at the input and output shafts. Based on obtained results, a promising, low-vibration cycloidal gearbox design with removed 7 lobes of the single wheel was studied using the FFT algorithm.