Application of relevance vector machine in the engine oil wear particle fault diagnosis

Diagnosis of engine fault is critical in reducing operating and maintenance costs which were paid more attention by many researchers. Due to the potential advantages to be gained from proactive maintenance, monitoring of machines health, assessing survival probability of machines unit. Numerous methods have been developed based on intelligent systems. This paper concerns with the prediction of engine oil wear particle using regression relevance vector machine(RVM). It attempts to overcome the problems of support vector machine (SVM) such as low sparsity, low computationally efficient and kernel function must be satisfied with the Mercer´s condition. RVM is a method suitable for processing regression and classification problems. In this paper, a regress prediction model is developed using RVM. The results of experiments show that RVM provides better prediction accuracy and generalization than SVM and artificial neural network (ANN). At the same conditions, RVM can be widely used in engine oil wear particle analysis and fault prediction.