Application of Data Mining for Determining Baselines of Wear Behavior in Engines, Using Oil Analysis Results

Although nowadays machinery oil analysis Condition Monitoring (CM) techniques are known as an effective method in abnormal wear in equipments and mechanical systems fault diagnosis; issues like wear behavior, technical features, and previous records of oil analysis results are essential and determinant in the process of interpreting the results of oil analysis in implementing CM programs.In this research, it is intended to justify the importance of historic data on oil analysis for fault detection. With the access to decent information sources, the wear behaviors of diesel engines are studied. Also, the relation between the final status of engine and selected features in oil analysis is analyzed. The dissertation and analysis of determining effective features in condition monitoring of equipments and their contribution, is the issue that has been studied through a Data Mining model. Selected indicators in oil analysis are Density, Silesia, PQ, and the amount of wearing metals i.e. Ferrum, Aluminum, Lead, Copper, and Tin. As the case study, data for the truck BENZ2628 are analyzed by Artificial Neural Network, Decision Tree, Data Visualization and the results are presented. The future study of this project will yield to an intelligent model for fault diagnosis and prognosis in aforementioned.