Remaining useful life estimation based on discriminating shapelet extraction

Abstract

In the Prognostics and Health Management (PHM) domain, estimating the remaining useful life (RUL) of critical machinery is a challenging task. Various research topics as data acquisition, fusion, diagnostics, prognostics and decision are involved in this domain. This paper presents an approach to estimate the Remaining Useful Life of equipment based on shapelet extraction. This approach makes use, in an offline step, of a history of run-to-failure data to extract discriminative rul-shapelets, i.e. patterns that are correlated with the RUL of the considered equipment. A library of rul-shapelets is hence extracted at this step. Then, in an online step, these rul-shapelets are compared to testing units and the ones that match these units are used to estimate their RULs. Therefore, the RUL estimation of a testing unit is based on patterns that have been selected for their high correlation with the RUL. This approach is hence different from classical similaritybased approaches that attempt to match complete testing units (or only late instants of testing units) with training ones to estimate the RUL. The performance of our approach is assessed with a case study on the remaining useful life estimation of turbofan engines and performance is compared with other similarity-based approaches.

Multidimensional prognostics for rotating machinery: A review

“…the most similar cases). 128 Those HIs with the highest similarities are subsequently used to predict the RUL.…”

Rotating Machine Prognostics Using System-Level Models

“…The methods to obtain the health indicator trajectories include logistic regression [59], weighted averaging methods [60], and flux-based methods [61]. If the un-processed data already capture the progression of the degradation process, the data can remain multi-dimensional [62]. Then, the monitoring data are converted into instances.…”

“…Figure 4 shows the general framework of similaritybased prognostic models. Because the similarity-based approaches use training data to construct instances (health indicator trajectories or multidimensional monitoring variables), they are compatible with algorithms that extract health indicators for RUL prediction [62]. The advantage of similarity-based approaches is that they can achieve satisfactory and accurate predictions when abundant data are collected from a variety of failure modes.…”

“…Examples are given below to set out how various similarity based models have been used for RUL prediction. [62] developed an RUL prediction technique that employs the Shapelet extraction process to extract failure patterns from multivariate sensory data obtained from a turbofan engine simulation program, C-MAPSS. The RUL was calculated as the weighted sum of the failure patterns that are highly corrected with the residual life.…”

Ensemble Recurrent Neural Network-Based Residual Useful Life Prognostics of Aircraft Engines

“…With the application of a large number of highly sophisticated engineering equipment, residual useful life (RUL) prediction becomes an indispensable technique to improve the reliability of equipment and make accurate maintenance and utilization strategy [1][2][3]. Based on the analysis of historical sensor data, the method might deeply mine the degradation process of equipment performance state, and achieve the accurate prediction of its RUL.…”