INTRODUCTION

The utmost priority for any process industry is safety. Once that is guaranteed, the next focus will be to reduce maintenance costs through accurate prediction of the behavior of all critical assets. The prediction, more often than not, is preceded by detection and diagnostics. Therefore, the three pillars are interconnected rather than each existing independently. So much has been done over the last two decades on developing efficient algorithms for the three aspects. However, very little is implemented in actual systems. The goal of this short course is (i) to introduce the state-of-the-art to field practitioners, (ii) to enlighten those interested in pursuing this field as researcher, and (iii) to provide ammunition to those already engaged in research in this area. Having a dashboard for real-time monitoring is one thing, but what is the point if we do not have the right AI engine to do the detection, diagnostics, and prediction. This short course is partly intended to provide insights into the classification of available methods, comparison and contrast between various methods, remarks for certain contexts and information on the big picture as it relates to IR 4.0. By and large, it cracks into the puzzle of what is best for advanced systems such as gas turbines, chillers, compressors, and antifriction bearings. The procedure for a scientifically acceptable selection of a particular algorithm for a particular application is one of the interesting takeaways from this short course.