PHYSICS-BASED DATA-DRIVEN METHODS FOR DIAGNOSTICS OF ROTATING MACHINERY

Abstract

Maintenance of rotating machinery is the essential part and core of every production process which directly affects its productivity and quality. As the level of complexity of modern rotating machinery grows, the need for an effective and efficient maintenance process increases as well. Based on numerous scientific papers in the field of condition monitoring, a conclusion can be drawn that the best indicator of the overall current state of the machine, which is sensitive to the appearance and development of a certain defect at the earliest stages is vibration. In order to accurately monitor and determine the current state of the machinery by measuring its vibrations, automation of the process of monitoring the condition and identification of possible faults has to be performed. The methods for automation generally can be divided into two large groups, physics-based and data-driven methods. This research gives an insight into the state-of-the-art of both methods and enhances their advantages and limitations. In order to overcome these limitations and take advantage of both, in the past few years a novel methodology in the field of rotating machinery diagnostics is proposed, the physics-based data-driven method. This research concentrates on discovering the potential for future work on physics-based data-driven methods in the field of rotating machinery. The first goal of this research would be the integration of purely physics-based models and purely data-driven models into a single hybrid model that would serve for vibration-diagnostic monitoring of the condition of rotating machines. In addition, the obtained from the hybrid model would be compared to the results obtained from the purely data-driven models, based on the obtained accuracy, the volume of the database, and calculation costs and time. In this way, it would be possible to conclude which method is superior for vibration-diagnostic monitoring of the condition of rotating machines.