A Change Point Detection Integrated Remaining Useful Life Estimation Model under Variable Operating Conditions

TL;DR

This work tackles the problem of accurate $RUL$ estimation under variable operating conditions by introducing a CVA-based, unsupervised change point detection framework that learns latent local temporal dynamics from normal operation. Detected change points are used to generate device-specific piecewise $RUL$ labels and to perform piecewise standardisation, enabling an LSTM to predict $RUL$ more reliably. The proposed offline-online pipeline demonstrates notable improvements on the C-MAPSS turbofan datasets, particularly for engines with multiple operating conditions, and highlights the importance of modelling heterogeneous change points. The approach is data-driven and does not rely on fixed domain knowledge, offering generalizability to other domains and potential for robust online monitoring and maintenance decision support.

Abstract

By informing the onset of the degradation process, health status evaluation serves as a significant preliminary step for reliable remaining useful life (RUL) estimation of complex equipment. This paper proposes a novel temporal dynamics learning-based model for detecting change points of individual devices, even under variable operating conditions, and utilises the learnt change points to improve the RUL estimation accuracy. During offline model development, the multivariate sensor data are decomposed to learn fused temporal correlation features that are generalisable and representative of normal operation dynamics across multiple operating conditions. Monitoring statistics and control limit thresholds for normal behaviour are dynamically constructed from these learnt temporal features for the unsupervised detection of device-level change points. The detected change points then inform the degradation data labelling for training a long short-term memory (LSTM)-based RUL estimation model. During online monitoring, the temporal correlation dynamics of a query device is monitored for breach of the control limit derived in offline training. If a change point is detected, the device's RUL is estimated with the well-trained offline model for early preventive action. Using C-MAPSS turbofan engines as the case study, the proposed method improved the accuracy by 5.6\% and 7.5\% for two scenarios with six operating conditions, when compared to existing LSTM-based RUL estimation models that do not consider heterogeneous change points.

Figures (10)

• Figure 1: Piecewise RUL function showing the relationship between the change point and upper RUL limit value, using randomly selected engines of FD004 in the Commercial Modular Aero-Propulsion System Simulation dataset.
• Figure 2: Overall structure of proposed change point detection-based RUL estimation method.
• Figure 3: Data segmentation using sliding window method with two randomly selected sensor signals shown as examples.
• Figure 4: Sample of uninformative sensor readings with constant values for randomly selected engines from FD001 and FD003, and with erratic, range-bound patterns for randomly selected engines from FD002 and FD004.
• Figure 5: Monitoring statistics, $T^2$ and $Q$ during normal operation with six operating conditions (a) first 60 operational cycles, and (b) next 20 operational cycles for Engine 116 of FD004.