The Proportional Hazards Model (PHM) under a Condition-Based Maintenance (CBM) policy is used by asset-intensive industries to predict failure rate, reliability function, and maintenance decisions based on vital covariates data. Cox’s partial likelihood optimization is a method to assess the weight of time and conditions into the hazard rate; however, parameter estimation with diverse covariates problem could have multiple and feasible solutions. Therefore, the boundary assessment and the initial value strategy are critical matters to consider. This paper analyzes innovative non/semi-parametric approaches to address this problem. Specifically, we incorporate IPCRidge for defining boundaries and use Gradient Boosting and Random Forest for estimating seed values for covariates weighting. When applied to a real case study, the integration of data scaling streamlines the handling of condition data with diverse orders of magnitude and units. This enhancement simplifies the modeling process and ensures a more comprehensive and accurate underlying data analysis. Finally, the proposed method shows an innovative path for assessing condition weights and Weibull parameters with data-driven approaches and advanced algorithms, increasing the robustness of non-convex log-likelihood optimization, and strengthening the PHM model with multiple covariates by easing its interpretation for predictive maintenance purposes.

Jointing Condition-Based Maintenance (CBM) with the Proportional Hazards Model (PHM), asset-intensive industries often monitor vital covariates to predict failure rate, the reliability function, and maintenance decisions. This analysis requires defining the transition probabilities of asset conditions evolving among states over time. When only one covariate is assessed, the model’s parameters are commonly obtained from expert opinions to provide state bands directly. However, the challenge lies within multiple covariate problems, where arbitrary judgment can be difficult and debatable, since the composite measurement does not represent any physical magnitude. In addition, selecting covariates lacks procedures to prioritize the most relevant ones. Therefore, the present work aimed to determine multiple covariate bands for the transition probability matrix via supervised classification and unsupervised clustering. We used Machine Learning (ML) to strengthen the PHM model and to complement expert knowledge. This paper allows obtaining the number of covariate bands and the optimal limits of each one when dealing with predictive maintenance decisions. This novel proposal of an ML condition assessment is a robust alternative to the expert criterion to provide accurate results, increasing the expectation of the remaining useful life for critical assets. Finally, this research has built an enriched bridge between the decision areas of predictive maintenance and Data Science.