Statistical learning for maintenance optimization: modeling the hazard function with variable recovery factors

Abstract

Abstract: This study aims to utilize advanced statistical methods in equipment maintenance within the machining and casting processes of an industrial company, focusing on cost optimization. Employing a methodology that combines literature review, mathematical formulation, and statistical treatment of historical data from preventive and corrective maintenance, the study stands out for the use of maximum likelihood estimation to determine not only the model parameters but also the recovery factor used during preventive maintenance. It also calculates the optimal maintenance time to minimize the total cost. This work not only underscores the importance of a data-driven maintenance approach, balancing reliability and cost to achieve effective optimization but also signals the need for future studies, especially in simulations to further enhance preventive maintenance scheduling, encompassing all aspects of the production process.