SELF ORGANIZING MAPS FOR SURVEYING LUBRICATION WITHIN SQUEEZE FILM DAMPERS.

Self-Organizing Maps (SOMs) represent a well known neural network technique particularly suited to classification tasks. Here it is adopted to monitor the lubricating conditions within squeeze film dampers for rotor support and was aimed, in particular, at identifying the aspect of the pressure waves within the oil film. Pressure distribution is indeed significantly influenced by a number of factors which affect damper operation. Results from past research in the field make it possible to infer that the pattern of pressure signals taken in the oil film represents a valuable source of information as regards the lubricating conditions within the damper. Surveillance procedures in the operation of turbomachinery could benefit from prompt detection of possible, unwanted changes in the characteristics of lubrication, for examples, when modeling bearing operations within model-based schemes. In this paper, SOM capabilities are first evaluated, dealing separately with theoretically simulated data. Subsequent tests adopted theoretical data as a reference for identifying experimental conditions. Further tests were carried out to map experimental data. Despite constraints consisting in the damper motion being imposed during the theoretical and experimental tests, the results confirmed the potential of the method and encourage further tests in conditions which are closer to real operation.