Abstract

In this webinar, we will talk about the advancements in fault detection and predictive maintenance for complex physical systems in the era of Industry 4.0. With the vast amount of data available today, there is a growing potential to effectively predict and prevent issues in industrial machinery. Multi-physics simulation models of reciprocating compressors as part of Digital Twin Environments have advanced significantly, thus allowing for the quick and easy generation of synthetic training datasets and specific labelled anomalous datasets simulating real-world faults. We will talk about a well-known commercial multi-physics simulation software that is used to model an industrial reciprocating compressor system with various domains including bearings, valves, compressor shaft mechanics, NVH, lubrication, and gas flow dynamics. The model is exercised with variability of typical physical parameters to simulate real-world operation of well-running, non-faulty reciprocating compressor systems and generate “no-fault” virtual input datasets. Faults are then intentionally introduced within the digital twin system to simulate real-world anomalies manifested via behavioral changes in signals of typical sensor data (e.g., compressed gas pressure) present in actual compressor systems. Supervised learning techniques are applied to train and optimize machine learning models for fault detection. The trained models can be deployed in real-world scenarios with online real-time monitoring to detect potential failures and identify fault causes. This can help with predictive maintenance of compressor systems, reduce unscheduled downtime, and thereby decrease the economic impact from loss of production. The promise of such technology is demonstrated as the ability to reproduce and prevent faults without actually creating them on a real machine, which would risk severely damaging these large and expensive machines.

Overall, this webinar will demonstrate the promise of technology in reproducing and preventing faults without risking damage to expensive machinery.

Topics include: