Development of Intelligent Techniques for Machine Prognostics

The prognostic system plays a crucial role in estimating the remaining useful life of machine components and forecasting of the future states of machines. The techniques related to prognostics consist of statistical-based, model-based, and data driven or intelligence-based. Among these, artificial intelligence is commonly used due to its flexibility in generating appropriate models for the forecasting purpose. This paper presents the development of intelligent techniques for machine health prognostic system in Intelligent Mechanics Laboratory (IML) of Pukyong National University (PKNU), South Korea. These developed techniques include support vector machine, relevance vector machine, Dempster-Shafer theory, decision tree, neuro-fuzzy inference systems. Additionally, they are also combined with other model-based techniques such as autoregressive moving average, proportional hazard model, logistic regression, etc. to fulfill the final goal of prognostic system. Case studies of machine health prognostics are also presented in this paper to show the plausibility of the developed systems.