Degradation trend feature generation improved rotating machines RUL prognosis method with limited run-to-failure data

Abstract

The success of data-driven remaining useful life (RUL) prognosis approaches of rotating machines depends heavily on the abundance of entire life cycle data. However, it is difficult to obtain sufficient run-to-failure data in industrial practice. Data generation technology is a promising solution for enriching data but fails to address the intrinsic complexity of non-linear stage degradation and the time correlation of long-term data. This research proposes an RUL prognosis approach improved by the degradation trend feature generation variational autoencoder (DTVAE). First, this study develops a framework combining degradation trend generation features to resolve the issue of capturing the elements of time distribution for run-to-failure data. Second, a generation variational autoencoder network with a tendency block is proposed to create high-quality correlation features of time series data. Third, original and created degradation trend features are subjected to deep adaptive fusion and health indicator extraction. A bi-directional long short-term memory network is employed to predict the degradation trend and obtain the RUL prognosis. Finally, the proposed approach’s feasibility is confirmed by cross-validation experiments on a bearing dataset, which reduces the prediction error by 22.309%.

Prescriptive maintenance for complex products with digital twin considering production planning and resource constraints

“…Xu et al [47] presented a dual-stream self-attention from another perspective, which is capable of learning features from both original and auxiliary data. Other methods such as autoencoder [48], deep belief networks [49,50], Transformer and its variants [51][52][53] have also been applied to RUL prediction successively.…”

Construction health indicator using physically-informed 1D-WGAN-GP joint attention LSTM-DenseNet method

“…Deep learning methods are typically employed for constructing various HI [25][26][27]. Zhang et al [28] presented a methodology that employs a variational autoencoder for the generation of degradation trend features in RUL forecasting for rotary machines, utilizing a bidirectional long short-term memory (Bi-LSTM) network to extract health indicator features. This addresses the inherent complexity of time-related dependencies in nonlinear degradation stages and long-term data; the method's feasibility has been verified through experiments.…”