Grant: Discovery Grants Program - Individual

Industry 4.0 has transformed many Canadian sectors, pushing them rapidly into digital and smart processes with high efficiency, safety, and sustainability requirements for industrial machinery. Machine condition monitoring (MCM) is crucial to achieving optimal maintenance and boosting equipment reliability and efficiency. Effective fault diagnosis and prognosis (FDP) can enable condition-based predictive maintenance for industrial assets where the correct maintenance action takes place at the correct time, minimizing system downtime, energy waste, and operational costs. With more and more connected and monitored equipment, data-driven approaches have emerged as the most promising tool for MCM. However, the growth of data-driven methods is nowhere near a reflection of their full potential due to the fundamental challenges including the difficulty in processing complex data with compound faults and heavy noise, limited data with labelled information, lack of data with fault conditions, and low model generosity and reliability due to varying working conditions. A revolutionary strategy is required to address this industry challenge, combining modelling, optimizations, and advanced artificial intelligence (AI). To enable the adaptability of intelligent FDP models in practical applications, this research program will develop original and innovative methodology combining both data-driven and physics-based models to ensure reliable and effective FDP for rotating machinery widely used throughout the Canadian manufacturing sector. By building a digital twin model of the physical system, simulation data mirroring the actual system is produced to extract useful information using deep learning models. The critical problem of prediction uncertainty will be estimated to achieve reliable FDP and adaptive model evolution. This research program will provide enriched training opportunities for a broad spectrum of HQP. Training HQP to move into impactful positions and produce quality results, is philosophy of the applicant's research program resulting in capable and enthusiastic graduates who make meaningful contributions to their fields. The proposed research will increase the number of transformative impacts in Canada's manufacturing sector, helping to enhance competitiveness, increase productivity and safety, while also reducing overall cost(s), energy consumption, and carbon emissions. This research program will address the most pressing industry challenges (e.g., lack of data, low reliability, and transferability) in achieving effective condition monitoring of industrial assets. The knowledge acquired from this program could also impact a wide range of industrial assets through developing MCM, including wind turbines, jet engines (airplanes), and other manufacturing machinery. Effective condition monitoring with advanced machine learning and cyber-physical systems can also impact condition monitoring tasks in transportation, healthcare, and civil engineering.