Scholarlink

<p>Health analysis and remaining useful life (RUL) estimation of turbofan engines are two essential components for accurate designing of proactive maintenance strategies that can prevent unexpected failures and minimize costly downtime. However, the performance degradation process of these engines follows a complex trajectory and involves multiple change points due to abrupt changes in operational or environmental conditions. This issue poses significant challenges for RUL estimation and prognostic strategy designing for turbofan engines. This issue is overlooked in most of the existing studies. This paper proposes a hybrid approach to address this challenge. Initially, a dynamic programming-based algorithm is proposed for the precise detection of change points that effectively captures the nonlinear and multistage nature of engine’s performance degradation. Based on the identified change points, a health indicator function is then designed to analyze the engine’s health condition. A Bayesian Long Short-Term Memory (LSTM) network-based model is proposed later by incorporating Monte Carlo dropout technique to account for both aleatory and epistemic uncertainties. This probabilistic RUL prediction approach provides more reliable and interpretable RUL estimates, which facilitate informed and risk-aware maintenance decisions. Finally, a multistage maintenance scheduling model is formulated by integrating change points and predicted RUL. The Genetic Algorithm has been designed to solve this optimization problem. This strategy optimizes maintenance costs, which reduces life-cycle costs while improving system availability. The proposed framework is validated on standard turbofan engine datasets and compared with state-of-the-art approaches. These results demonstrate superior performance in RUL prediction for the proposed approach. This study offers significant practical contributions to prognostics and maintenance engineering, with important implications for enhancing the safety, efficiency, and reliability of aerospace systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Remaining useful life prediction and maintenance schedule designing for turbofan engines considering multistage change points and uncertainty awareness

  • 期刊论文

  • 发表日期：

    2026年6月26日

查看全文
  • Deepjyoti Saha

摘要

Health analysis and remaining useful life (RUL) estimation of turbofan engines are two essential components for accurate designing of proactive maintenance strategies that can prevent unexpected failures and minimize costly downtime. However, the performance degradation process of these engines follows a complex trajectory and involves multiple change points due to abrupt changes in operational or environmental conditions. This issue poses significant challenges for RUL estimation and prognostic strategy designing for turbofan engines. This issue is overlooked in most of the existing studies. This paper proposes a hybrid approach to address this challenge. Initially, a dynamic programming-based algorithm is proposed for the precise detection of change points that effectively captures the nonlinear and multistage nature of engine’s performance degradation. Based on the identified change points, a health indicator function is then designed to analyze the engine’s health condition. A Bayesian Long Short-Term Memory (LSTM) network-based model is proposed later by incorporating Monte Carlo dropout technique to account for both aleatory and epistemic uncertainties. This probabilistic RUL prediction approach provides more reliable and interpretable RUL estimates, which facilitate informed and risk-aware maintenance decisions. Finally, a multistage maintenance scheduling model is formulated by integrating change points and predicted RUL. The Genetic Algorithm has been designed to solve this optimization problem. This strategy optimizes maintenance costs, which reduces life-cycle costs while improving system availability. The proposed framework is validated on standard turbofan engine datasets and compared with state-of-the-art approaches. These results demonstrate superior performance in RUL prediction for the proposed approach. This study offers significant practical contributions to prognostics and maintenance engineering, with important implications for enhancing the safety, efficiency, and reliability of aerospace systems.