Enhancing Asset Reliability and Sustainability: A Comparative Study of Neural Networks and ARIMAX in Predictive Maintenance

Organizations strive to maximize efficiency in their manufacturing processes, yet they must also consider broader repercussions, as industrial activity directly impacts the environment and society. The adoption of innovative technologies and initiatives to mitigate this impact is therefore essential. Traditional asset maintenance plays a critical role in ensuring high equipment availability, but there is a clear need to evolve toward predictive and sustainable maintenance strategies to enhance reliability, safety, and equipment lifespan. This shift redefines maintenance itself, aligning it with Circular Economy principles and Industry 4.0 solutions to prevent unplanned downtime, reduce failures, and improve personnel and facility safety. This research examines the transition from traditional preventive maintenance to predictive and sustainable maintenance in a real-world industrial context, comparing the design of a neural network with the ARIMAX technique to develop reliable predictive models. The study aims to facilitate a paradigm shift by proposing a predictive model that reduces unplanned shutdowns and optimizes spare parts and labor utilization. The practical application focuses on a hydrogen compressor in the petrochemical industry, demonstrating the model’s potential for operational and sustainability improvements.