Transfer Learning For Predictive Maintenance: A Case Study

In light of recent strides in high-performance computing, the concept of transfer learning has emerged as a prominent paradigm within the realm of Artificial Intelligence and Machine Learning methodologies. Analogous to the human brain's capacity to assimilate information across related domains for pattern recognition, transfer learning has swiftly asserted its dominance, particularly in deep learning applications such as image classification and natural language processing. Despite its ascendancy in these domains, there exists a lack of comprehensive investigations in alternative domains, notably those encompassing tabular data formats. This thesis seeks to redress this gap by conducting an empirical examination of transfer learning within the context of predictive maintenance. The study employs a case study methodology involving induced misalignments in various levels of horsepower motors. A comparative analysis is undertaken, juxtaposing transfer learning techniques against standard deep learning approaches, with the aim of elucidating the overarching efficacy and viability of transfer learning within the domain of predictive maintenance. This empirical exploration contributes to the broader discourse on the applicability and optimization of transfer learning methodologies across diverse domains, specifically addressing its efficacy in the nuanced context of predictive maintenance for machinery.