Information and modern computing technology advancements have led to a rise in the importance of maintenance, particularly in areas where a single components failure could have a significant impact on the overall systems performance. Numerous industries, including Alfa Laval, are operating on conditional-based systems that provide warnings only when a machine fails. In the worst instances, pro- longed downtime or machine failure can be costly in terms of money, time, and security [2]. The Alfa Laval company is interested in build- ing a smart alarm system that anticipates alarms and warnings based on sensor readings. For solving these issues, predictive maintenance using machine learning is one of the most effective approach to de- tect the machine condition in advance for maintenance and prevent it from real-time damage or faults. To obtain the best prescient machine learning model, we examined multi-linear and non-linear methods with tensor representation and the linear method as a baseline on real-time multi-sensor time-series datasets to build the smart alarm predictive system to anticipate cautions and warnings. As per the ex- perimental results, we are more certain that the non-linear (Tensor Convolutional Neural Network) method is more ideal than the other methods for the company’s multivariate time series datasets.
Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:hh-46193 |
Date | January 2022 |
Creators | Sahu, Chandraprakash, Buhari, Ahamed |
Publisher | Högskolan i Halmstad, Akademin för informationsteknologi |
Source Sets | DiVA Archive at Upsalla University |
Language | English |
Detected Language | English |
Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
Format | application/pdf |
Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.0021 seconds