Mechanical elements, gears and rolling bearings susceptible to wear and failures. This is because of the relative motion between the machine elements. As such various techniques like temperature monitoring and chemical content analysis have been developed to analyze and monitor the condition of machine bearings. However, vibration analysis is the main method used despite the noise and disturbances. Therefore, researchers have developed numerous algorithms with the aim of achieving clear signals, by eliminating the vibration disturbances, to monitor the health conditions of bearings. They provide both real-time measurements and simulated results. Unfortunately, unlike the simulated results which are readily available as open source for post-processing, real measurements results and not that common.
Previously epicyclic gear-like model was proposed to determine the faulted bearing simulation signals. It enabled the evaluation of the fault frequency characteristics resulting from damaged bearings. Owing to the limitations of the method like the slippery effect occurring at the contacts of the bearings and other components, various improvements have been initiated to improve their efficiency. For instance, the vibration signal model have been modeled as a cyclostationary signal with a periodic autocorrection function. It provides a better description of the slippery effects which was the main drawback of the initial models. Consequently, recently developed models have also taken into consideration the load and speed variations in the machine conditions thus improving their efficiency and reliability. Unfortunately, due to the complex nature of the models, the algorithms have become more complex hence difficult to implement.
Dr. Gianluca D’Elia and Professor Emiliano Mucchi at the University of Ferrara in collaboration with Dr. Marco Cocconcelli at University of Modena and Reggio Emilia developed a novel algorithm to simulate the vibration signals of a faulted bearing. The model was an improvement of the model proposed by Prof. Jerome Antoni (University of Lyon), including extending the constant speed operation to variable speed and developing new mathematical formulation. The simulation is an open source high-level language similar to Matlab. The authors’ main aim was to develop a universal model acceptable to the researchers’ community for conditions monitoring. Their work is published in the journal, Meccanica.
The authors observed that the developed model could be used by various readers and researchers to simulate various bearing faults and operating conditions because it is based on an open-source algorithm. Consequently, a similar resonance frequency was recorded for both the stimulation signal and the experimental data. All the expected feature and additional improvements were effectively analyzed and proved.
The developed novel model has successfully addressed the challenges of the initial models. For example, it allows the user to set various features including selection of the fault location, selection of the fault stage, signal cyclostationarity, deterministic contributions, random contributions, the machine working condition either as stationary or non-stationary and the effect of the resonances. Therefore, the study is a foundation for the realization of universally accepted tools for condition monitoring and diagnosis techniques in machines.
D’Elia, G., Cocconcelli, M., & Mucchi, E. (2017). An algorithm for the simulation of faulted bearings in non-stationary conditions. Meccanica, 53(4-5), 1147-1166.Go To Meccanica