An improved Cox proportional hazards model for reliability analysis of aviation gas turbines considering varying environmental conditions and operational settings

A reliable aviation gas turbine reduces maintenance costs and improves efficiency in the aerospace industry. The Cox proportional hazards model is widely used in reliability analysis to study the effect of multiple covariates on failure. Recent studies combine this model with machine learning to enhance gas turbine reliability analysis, often using environmental conditions, operational settings, and gas path parameters as covariates. However, these direct covariates may not fully capture gas turbine degradation due to the effects of fluctuating environmental conditions and operational settings. To address this issue, this paper proposes a method that employs machine learning to develop new covariates for the Cox proportional hazards model. Machine learning models a healthy turbine, generating covariates less affected by varying environmental conditions and operational settings, better representing gas turbine degradation. This approach addresses limitations in using traditional covariates and enhances the Cox proportional hazards model's ability to analyze reliability under varying working conditions. The method is validated with data from NASA’s C-MAPSS dataset, showing improved accuracy compared to the traditional Cox proportional hazards model.