A COMPARATIVE EVALUATION OF SUPERVISED MACHINE LEARNING METHODS TO PREDICT MACHINERY CONDITION AND RELIABILITY USING LUBRICATION PARAMETERS

Abstract

Effective machinery lubrication is essential for ensuring the reliability and availability of systems, as it plays a vital role in the smooth operation of machines, much like blood in the human body. Inadequate lubrication and poor management practices can have severe consequences, including equipment failures, compromised safety, and increased maintenance costs, particularly in asset-intensive industries such as oil, gas, and petrochemical facilities. To address this challenge, this study utilizes artificial intelligence, machine learning, and data mining techniques to develop and compare various predictive models for monitoring machine health. The ultimate goal of this research is to identify the most effective approach for predicting machine health using lubrication condition data. To achieve this, relevant lubrication features were carefully selected and incorporated into model development, with their significance in predicting machine health subsequently evaluated. An analysis of real-world lubrication data from diverse machines revealed that the random forest algorithm outperformed other algorithms in meeting our requirements. The findings of this study demonstrate the potential of lubrication data to forecast machine health and improve equipment reliability, providing valuable insights that enable researchers to enhance lubrication management strategies. Overall, this research contributes to the development of enhanced lubrication management strategies, which can help reduce downtime, promote more efficient and safe operations, and ultimately benefit asset-intensive industries.