- Available Technology
This failure methodology utilizes purposeful generation of labeled failure mode datasets via surrogate purposeful failure twin (SPFT) simulations. Machine learning powered performance analysis, predictive maintenance, and production management capable of reducing unexpected failure and unnecessary downtime are enabled via this innovation. This technology revolutionizes the manufacturing operations management (MOM) system market, which currently suffers from a gap in the availability of diagnostic and predictive failure data for manufacturing equipment. Current failure data generation methodologies are time consuming and costly, or do not represent all possible failure modes experienced by the equipment. Failing equipment can lead to safety hazards, reduced product quality, and avoidable reduction in output capacity. Improving failure metrics, real-time analysis, and incorporating artificial intelligence methodologies can alleviate these negative occurrences and bring manufacturers into the Industry 4.0 future with significantly improved overall equipment effectiveness (OEE). Clemson University researchers have developed a Purposeful Failure Methodology, which generates labeled relevant datasets via SPFT simulation and utilizes machine learning to not only improve prescriptive equipment maintenance, but also pinpoint equipment failure modes through real-time data collection from in-line equipment.
Failure Mode Methodologies, Manufacturing, Equipment, Learning Algorithms, Dataset Analysis
The Purposeful Failure Methodology standardizes failure data generation. This enables machine learning backed diagnosis and prognosis of equipment failure modes. The methodology utilizes intentional component and system damage growth from identified points of failure that is tracked by commonplace sensors. The generated data is labeled prior to input into machine learning algorithms, which are then utilized to inform diagnostic and prognostic systems. The methodology reproduces operational component environments during the generation of failure data via SPFT models leading to direct in-line equipment monitoring and failure prediction translatability. The nature of the methodology is conducive for rapid labeled failure dataset generation, equipment monitoring, and informed predictive maintenance scheduling. This provides potential cost and downtime savings, increased production, and improved employee safety.
Stay up-to-date with the latest trends in the innovation and research industry. Sign up for our newsletter to see how CURF is making a difference and impacting the economy where we live.