Hybrid Modeling For Asset Performance
Peter DamerSolutions Architect - ABB Energy Industries
Time & Location
Thursday: 10.00 to 10.20, Stage 4
About this presentation
The use of mathematical models has, for many years, been key in understanding equipment condition and the early detection of developing faults. These have been traditionally based on either physical principles or “word” models such as Failure Modes and Effects or Root Cause Analysis.
The last few years has seen a meteoric rise in the interest in statistical and other data driven techniques where machine learning is used to improve the performance of online asset monitoring, and provide predictions of future equipment condition.
The current dominance of AI in all areas of science and engineering tends to eclipse the value of more established techniques. In this presentation, the case is made for a hybrid approach, where the “sweet spot” of performance is achieved through the considered adoption and integration of a number of modeling approaches.
It explains how machine learning is used to unlock the features present in equipment data without replacing the complementary intelligence and experience of the asset engineer.
Peter brings over thirty years’ experience in industrial control systems design and operation to his current role as the Global Solutions Architect for asset performance.
He has worked as a design and commissioning engineer across discrete, batch and continuous processing sectors. With a first degree in Engineering Science and a Masters in Process Automation, his expertise includes Manufacturing Execution Systems, advanced control and software engineering.
This broad experience provides the expertise necessary to apply both traditional and emerging techniques to a range of industrial applications.
Peter also sits on the IChemE Process Management and Control Committee and is Chair of the Process Automation Apprenticeship Group