Abstract
Remaining useful life (RUL) estimation is expected to provide appropriate maintenance for components or systems in industry to improve the reliability of the systems. Most data-based methods are limited to a single model, which is susceptible to various factors like environmental variability and diversity of operating conditions. In this paper, we propose an optimal stacking ensemble method combining different learning algorithms as meta-learners to mitigate the impact of multi-operating conditions. The selection of meta-learners follows a multi-objective evolutionary algorithm named non-dominated sorting genetic algorithms-II to balance the two conflicting objectives in terms of accuracy and diversity. Then the eventually evolved meta-learners are integrated by the meta-classifier for RUL estimation. In addition, a long-short-term feature extraction strategy is proposed to capture more degradation information from lifecycle data dynamically. Extensive experiments are performed on aero-engine dataset and battery dataset provided by NASA, which achieves the higher prognostic accuracy compared with the single models and existing methods.
Highlights
Intelligent prognostic and health management (PHM) has been widely used in aircrafts, automobiles, nuclear power plants, dams, and other industrial fields [1]. It comprehensively utilizes the latest research of modern information and artificial intelligence technologies to provide a powerful tool to health management, which is significant in industries [2]
Consider the range fluctuation in time series data caused by unstable environmental variability [19], diversity of operating modes (e.g.,loading, usage, rotatory speed) [20], nonlinear degradation modes with noise [21], it is difficult to use a single model mapping from signal features to Remaining useful life (RUL) values in a multi-noise and time-varying environment [22]
In this paper, the optimal stacking method is proposed for the RUL prediction of engineered systems by evolving the most suitable meta-learners simultaneously subject to accuracy and diversity as two conflicting objectives
Summary
Intelligent prognostic and health management (PHM) has been widely used in aircrafts, automobiles, nuclear power plants, dams, and other industrial fields [1]. Consider the range fluctuation in time series data caused by unstable environmental variability (e.g., temperature, relative humidity, pressure) [19], diversity of operating modes (e.g.,loading, usage, rotatory speed) [20], nonlinear degradation modes with noise [21], it is difficult to use a single model mapping from signal features to RUL values in a multi-noise and time-varying environment [22]. The single-objective optimization algorithms usually adopt a greedy search strategy that leads to local minimum It doesn’t take much accuracy improvement but excess meta-learners. Literature [29] adopted a multi-objective optimization algorithm named non-dominated sorting genetic algorithms-II (NSGA-II) to evolve a ensemble and the result is averaged by each individual It maximizes the generalization capacity of the ensemble and minimize its structural complexity simultaneously to get a better ensemble.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
