Search

Abstract: In order to discriminate small changes for early fault diagnosis of rotating machines, condition monitoring demands that the measurement of instantaneous angular speed (IAS) of the machines be as accurate as possible. This paper develops the theoretical basis and practical implementation of IAS data acquisition and IAS estimation when noise influence is included. IAS data is modelled as a frequency modulated signal of which the signal-to-noise ratio can be improved by using a high-resolution encoder. From this signal model and analysis, optimal configurations for IAS data collection are addressed for high accuracy IAS measurement. Simultaneously, a method based on analytic signal concept and fast Fourier transform is also developed for efficient and accurate estimation of IAS. Finally, a fault diagnosis is carried out on an electric induction motor driving system using IAS measurement. The diagnosis results show that using a high-resolution encoder and a long data stream can achieve noise reduction by more than 10dB in the frequency range of interest, validating the model and algorithm developed. Moreover, the results demonstrate that IAS measurement outperforms conventional vibration in diagnosis of incipient faults of motor rotor bar defects and shaft misalignment. [Copyright &y& Elsevier]

• A full generalization of Gini index is proposed for sparse quantification. • Fully generalized Gini indices are theoretically proven to be sparsity measures. • Two key performances of fully generalized Gini indices are investigated. • Fully generalized Gini indices exhibit excellent performance in characterizing repetitive transients. • Fully generalized Gini indices can effectively achieve bearing condition monitoring and fault diagnosis. The classic Gini index (GI) is generalized recently by using nonlinear weight sequences as sparsity measures for sparse quantification and machine condition monitoring. The generalized GIs with different weight parameters are more robust to random transients. However, they show insufficient performance in discriminating repetitive transients under noise contamination. To overcome this shortage, this paper proposes a two-parameter generalization method to tune not only the weight parameter but also the norm order, allowing for a full generalization of the classic GI to quantify transient features and leading to new statistical indicators which are named fully generalized GIs (FGGIs). Mathematical derivations show that FGGIs satisfy at least four of the six typical attributes of sparsity measures and that those with weight parameter equal to one satisfy at least five sparse attributes, proving that they are a new family of sparsity measures. Numerical simulations demonstrate that FGGIs can monotonically evaluate the sparseness of the signals and that the FGGIs with appropriate parameters exhibit improved performance in resisting random transient interferences and discriminating noise-contaminated repetitive transients compared to traditional sparsity measures. The performance of FGGIs in the condition monitoring of rolling element bearings is validated using two different run-to-failure experiment datasets, including a gradual failure and a sudden failure. The results show that increasing the norm order can improve the capability of FGGIs to characterize transient fault features, allowing more accurate trending of bearing health conditions, and therefore achieving better condition monitoring performance than the traditional sparsity measures. [ABSTRACT FROM AUTHOR]

• A multiscale denoising branch is developed to extract multi-level information and reduce noise interference. • An improved flow direction strategy-based adaptive resonance branch is introduced to learn fault-related periodic impulsive features. • Case studies are conducted to validate the efficacy of the developed IFD-MDCN. Rolling bearings are the core components of rotating machinery, and their normal operation is crucial to the entire industrial production. Most existing condition monitoring methods have been devoted to extracting discriminative features from vibration signals that reflect bearing health status information. However, the complex working conditions of rolling bearings often make the periodic impulsive characteristics related to fault information easily buried in noise interferences. Therefore, it is challenging for existing approaches to learning discriminative fault-related features in these scenarios. To address this issue, a novel multibranch CNN named IFD-MDCN is developed in this paper, which represents multibranch denoising convolutional networks (MDCN) with an improved flow direction (IFD) strategy. The main contributions of this work include: (1) designing a multiscale denoising branch to extract multi-level information and reduce noise impact. More specifically, the multiscale denoising branch adopts a Gaussian multi-level noise reduction procedure to represent vibration signals at multiple levels and filter out the noise components, and then it uses a multiscale convolutional module to extract abundant features from these denoised signal representations; (2) establishing an improved flow direction strategy-based adaptive resonance branch to learn periodic impulsive features associated with fault information from vibration signals. Extensive experimental results reveal that the IFD-MDCN outperforms five state-of-the-art approaches, especially in strong noise scenarios. [ABSTRACT FROM AUTHOR]

• An improved form of SWNSE is developed by generalizing the order of the norm. • A design framework for improved Gini indices is established and four specific forms of improved Gini index are proposed. • Improved Gini indices are proven to be sparsity measures. • A series of improved Gini indices are proposed for characterizing repetitive transients. • The improved Gini indices with norm order 3 exhibit better comprehensive performance. Sparsity measures are important and effective tools for accurately characterizing fault features and degradation trends of rotating machinery. In the past few decades, numerous sparsity measures, such as kurtosis, negentropy, Lp/Lq norm, pq-mean, smoothness index and Gini index, have been proposed and introduced for condition monitoring, fault diagnosis and remaining life prediction of rotating machinery. However, it is difficult for traditional sparsity measures to possess strong random impulse resistance and fault impulse discernibility simultaneously. To design robust sparsity measures for repetitive transient characterization and machine condition monitoring, an improved form of the sum of weighted normalized square envelope (SWNSE) is firstly developed by generalizing the order of the norm as a framework for adaptively quantifying the repetitive transients. Under this framework, a rank-dependent generalized Gini index (RDGGI) is proposed, allowing four specific forms of RDGGI to be reasonably constructed by introducing four weight sequence design methods. Theoretical derivation shows that the quantitative indices produced by the four specific forms of RDGGI satisfy at least five of the six attributes of sparsity measures, proving that the newly designed quantitative indices are sparsity measures, and are also scale-invariant and have a limited magnitude range between 0 and 1. Furthermore, a series of improved Gini indices are derived by setting typical norm orders and designing appropriate monotonically decreasing weight sequences, and simulation analysis is conducted to verify their three key properties in characterizing transient features. Finally, the performance of the improved Gini indices in characterizing bearing fault features for condition monitoring is verified using vibration datasets from two bearing run-to-failure experiments. Simulation and experimental results show that the improved Gini indices with norm order 3 can simultaneously possess strong resistance to random impulses and discernibility of fault impulses, demonstrating the superior performance of improved Gini indices in bearing condition monitoring. [ABSTRACT FROM AUTHOR]