Combined Anomalies Prediction Using the Bayesian Theory


Hadjadj Aoul Elias, Electromechanical Department, Electromechanical System Laboratory, University Badji Mokhtar Annaba, Algeria.



The industry world uses machines and plants that are increasingly powerful and complex. The requirements of high safety, the reduction of the exploitation costs, and the control of the equipment availability give to maintenance a dominating role. The industrialists attach a great importance to the conditional maintenance of the revolving machines that use primarily the vibrations of their rotors. The temperature measurements in the stages can bring additional information to vibrations.

In conditional preventive maintenance, the diagnosis of failures of the industrial systems, if it is carried out with effectiveness, represents one of the means to gain points of productivity. It consists in observing the symptoms of a failure and then identifying the cause using a logical reasoning founded on observations, that is, to dismount a deterministic mechanism between the cause and its effect.

This research presents and discusses the decision making that is practically exerted with each stage in the procedure of industrial diagnosis and tool of assistance to the decision making. The approach used is Bayesian theory to reveal a defect masked by another in the low frequency (combined defect). The turbo compressor as an object of research, vibratory analyses, and thermography are the techniques used in this work. Copyright © 2011 John Wiley & Sons, Ltd.