Author
M. Elbadaoui
Bio: M. Elbadaoui is an academic researcher. The author has contributed to research in topics: Cyclostationary process & Computer science. The author has an hindex of 1, co-authored 1 publications receiving 66 citations.
Papers
More filters
TL;DR: This paper investigates and compares two emerging approaches to vibration-based fault detection based on a cyclostationary modeling of the bearing signal and addresses the extension of these approaches to the nonstationary operating regime.
101 citations
TL;DR: In this paper , two extensions of the synchronous average, namely the global and the local synchronous fitting, are studied. But the applicability of this technique is confined to the case where signals are recorded under quasi-stationary regimes, which is restrictive for many applications like aeronautics.
4 citations
TL;DR: In this paper , the authors proposed a vibration-based diagnostic methodology for aircraft bearings based on a joint first and second-order cyclostationary analysis, instead of ignoring the later or performing the diagnosis on an arbitrary stationary speed.
Abstract:
This paper proposes a vibration-based diagnostic methodology for aircraft bearings based on a joint first- and second-order cyclostationary analysis. The idea is to track the first- and second-order content over a predefined operating speed range, instead of ignoring the later or performing the diagnosis on an arbitrary stationary speed. The methodology applies to relatively long vibration signals recorded under strong speed variations, using a sliding window over which fluctuations are low. First, we obtain the time-evolution of the spectral statistics by computing the so-called instantaneous power and coherence spectra reflecting the first and second order content, respectively. Then, we design a time-to-speed transform based on fuzzy logic to transform the previously obtained time-cyclic maps into speed-cyclic maps, expressing the spectral statistics as functions of a predefined operating speed grid of interest. Last, we demonstrate the proposed methodology on a real vibration signal captured from an accessory gearbox of a CFM56 aircraft engine, with multiple bearing faults.
Cited by
More filters
TL;DR: An overview of the Deep Digital Twin framework is presented and it is demonstrated that the DDT is able to detect incipient faults, track asset degradation and differentiate between failure modes in both stationary and non-stationary operating conditions when trained on only healthy operating data.
122 citations
01 Jul 2020
TL;DR: The experimental results demonstrate that the proposed approach correctly identifies single and recurrent anomalies without any prior knowledge of their characteristics, outperforming by a large margin several competing approaches in accuracy, while being up to orders of magnitude faster.
Abstract: Subsequence anomaly detection in long sequences is an important problem with applications in a wide range of domains. However, the approaches that have been proposed so far in the literature have severe limitations: they either require prior domain knowledge that is used to design the anomaly discovery algorithms, or become cumbersome and expensive to use in situations with recurrent anomalies of the same type. In this work, we address these problems, and propose an unsupervised method suitable for domain agnostic subsequence anomaly detection. Our method, Series2Graph, is based on a graph representation of a novel low-dimensionality embedding of subsequences. Series2Graph needs neither labeled instances (like supervised techniques), nor anomaly-free data (like zero-positive learning techniques), and identifies anomalies of varying lengths. The experimental results, on the largest set of synthetic and real datasets used to date, demonstrate that the proposed approach correctly identifies single and recurrent anomalies without any prior knowledge of their characteristics, outperforming by a large margin several competing approaches in accuracy, while being up to orders of magnitude faster.
76 citations
TL;DR: A new transfer learning method based on bidirectional Gated Recurrent Unit (TBiGRU) is proposed to accurately predict the RUL of bearings under different working conditions and can adaptively recognize different running states of bearings and obtain corresponding training labels, and realize better RUL prediction performance under differentWorking conditions.
74 citations
TL;DR: In this article, a physics-informed deep learning approach was proposed for bearing condition monitoring and fault detection, which consists of a simple threshold model and a deep convolutional neural network (CNN) model.
56 citations
20 Apr 2020
TL;DR: NorM is a novel approach, suitable for domain-agnostic anomaly detection, based on a new data series primitive, which permits to detect anomalies based on their (dis)similarity to a model that represents normal behavior.
Abstract: Subsequence anomaly (or outlier) detection in long sequences is an important problem with applications in a wide range of domains. However, current approaches have severe limitations: they either require prior domain knowledge, or become cumbersome and expensive to use in situations with recurrent anomalies of the same type. In this work, we address these problems, and propose NorM, a novel approach, suitable for domain-agnostic anomaly detection. NorM is based on a new data series primitive, which permits to detect anomalies based on their (dis)similarity to a model that represents normal behavior. The experimental results on several real datasets demonstrate that the proposed approach outperforms by a large margin the current state-of-the art algorithms in terms of accuracy, while being orders of magnitude faster.
53 citations