Salvatore Manconi

A Machine Learning Approach to Fault Diagnosis of Rolling Bearings

This paper presents a method based on classification techniques for automatic fault diagnosis of rolling element bearings. Experimental results achieved on vibration signals collected by an accelerometer on an experimental test rig show that the method can automatically detect different types of faults. Furthermore, the method is able, once trained on an appropriate representative set of basic faults, to recognize more serious faults, provided they are of the same type. We also analyzed the trend of correct classification of bearing faults on variation of the signal-to-noise ratio achieving high levels of robustness.

An Experimental Investigation on Aerospace Quality Gears Operating in Loss of Lubrication Condition

This paper shows the results of an experimental study carried out on spur gears for aerospace applications operating in loss of lubrication. The aim of this work was to establish a baseline for gear behavior under oil off conditions. A total amount of 40 tests were performed with gears made from 2 materials operating at different levels of sliding speed and contact pressure. In some cases the bulk temperature was measured to evaluate the heating of the running gear. A more relevant wear and heating of the gears was observed with the increase of the contact pressure rather than with the increase of the sliding speed. In all the tests the transmission was able to transfer the required power. However some tests were stopped before the required time because of the overcoming of the rig safety threshold. The results showed a different oil off time depending on the material.Copyright

Experimental Study on Wear and Fracture in Aeronautical Gear Transmissions

This work presents some of the results obtained from an extensive experimental test campaign performed at the Department of Mechanical Nuclear and Production Engineering (DIMNP) of Pisa University with the partnership of Avio Propulsione Aerospaziale S.p.A. This test campaign, that is still being performed, is concerned with the characterization of teeth damage in high performance gear transmissions, such as aeronautical ones, characterized by high levels of velocity and load. Damage detection and monitoring have been pursued during all the tests in order to achieve a better understanding of causes and evolution of teeth damage. Moreover, to investigate the effects of different parameters on teeth damage, tests with different load, velocity, teeth geometry, materials and surface finishing, have been performed. The surface conditions of the gear teeth have been investigated using in particular two different instruments: a computer-controlled stylus profilometer able to perform 3D investigations and a scanning electron microscope (SEM) recently provided by a software for surface reconstruction. Qualitative and quantitative changes in surface roughness have been detected. Besides the experimental activity, numerical analyses have been also started for finding theoretical criteria for surface damage prediction; some few obtained results are presented. Due to the vastness of the work performed, only a survey of the main results obtained till now is presented in the paper.Copyright

Vibration Signature Analysis of Real and Virtual Damaged Gears

A high performance rig designed to test gears (materials and geometries) in extreme load and speed conditions as in aerospace applications was used to test diagnostic systems by the only vibration analysis. The authors of the present report have implemented, in addition to some standard signal analysis techniques, some advanced methodologies (e.g. NA4, NB4), which are applied to the synchronously averaged signal to detect the smallest fault related variations in the tested gear vibration signature. The different techniques have been tested and compared on the signals of a gear that underwent failure, showing the good performance of the latter approaches compared to the more conventional ones. A numerical simulation of the gear meshing was carried out to support the experimental campaign and the benchmarking of the diagnostic tools. The stiffness computed by a FEM code was employed to simulate the gear system dynamic behavior with a lumped model, which was finally tuned using the experimental signals, to obtain indications on the crack evolution.Copyright