A New Bearing Fault Diagnosis Method Based on Fine-to-Coarse Multiscale Permutation Entropy, Laplacian Score and SVM

Abstract

Fault diagnosis of rotating machinery is vital to identify incipient failures and avoid unexpected downtime in industrial systems. This paper proposes a new rolling bearing fault diagnosis method by integrating the fine-to-coarse multiscale permutation entropy (F2CMPE), Laplacian score (LS) and support vector machine (SVM). A novel entropy measure, named F2CMPE, was proposed by calculating permutation entropy via multiple-scale fine-grained and coarse-grained signals based on the wavelet packet decomposition. The entropy measure estimates the complexity of time series from both low-and highfrequency components. Moreover, the F2CMPE mitigates the drawback of producing time series with sharply reduced data length via the coarse-grained procedure in the conventional composite multiscale permutation entropy (CMPE). The comparative performance of the F2CMPE and CMPE is investigated by analyzing the synthetic and experimental signals for entropy-based feature extraction. In the proposed bearing fault diagnosis method, the F2CMPE is first used to extract the entropy-based features from bearing vibration signals. Then, LS and SVM are used for selection of features and fault classification, respectively. Finally, the effectiveness of the proposed method is verified for rolling bearing fault diagnosis using experimental vibration data sets, and the results have demonstrated the capability of the proposed method to recognize and identify the bearing fault patterns under different fault states and severity levels.INDEX TERMS Fault detection and diagnosis, fine-to-coarse multiscale permutation entropy, Laplacian score, support vector machine.

A New Fuzzy Logic Classifier Based on Multiscale Permutation Entropy and Its Application in Bearing Fault Diagnosis

“…The parameters of multiscale permutation entropy selected in this paper are an embedded dimension of 6 m  and delay time of 1   . In order to obtain the signal features as much as possible, the scale factor is set as 32 s  [33,34]. It can be seen that such a feature set has many scales and the entropy values are crossed together, which is not conducive to the final classification.…”

Entropy Measures in Machine Fault Diagnosis: Insights and Applications

“…For instance, a hierarchical decomposition is used in [78], preserving the strength of the multiscale decomposition with additional components of higher frequency in different scales. A fine-to-coarse procedure is developed in [79], aiming to generate multiple-scale components with fine-grained low-and high-frequency information, and to yield better consistent entropy values even with high scales and strong noise. • Multivariate analysis based entropy approaches: the complexity of multichannel data is assessed with multivariate extensions of MSEn where multichannel data is analyzed with a definition of multivariate single-scale entropy algorithm.…”

“…Further, some modified entropy measures have been proposed based on enhanced scale-extraction mechanisms. Related studies include generalized composite MPE [103], hierarchical entropy [104], modified hierarchical PerEn [105], and fine-to-coarse MPE [79]. In general, these methods earn higher consistency and reduced bias in time series complexity analysis, as compared with single-scale methods.…”

Related Entropy Theories Application in Condition Monitoring of Rotating Machineries

“…Zhang et al [70] singular value decomposition + permutation entropy 8 Wang et al [71] wavelet packet transform + permutation entropy 9 Zhao et al [72] wavelet packet decomposition + multiscale permutation entropy Fu et al [73] variational mode decomposition + permutation entropy Yan et al [74] improved variational mode decomposition + instantaneous energy distribution-permutation entropy Yasir et al [75] multi-scale permutation entropy Tian et al [76] permutation entropy + manifold-based dynamic time warping Lv et al [77] permutation entropy Zheng et al [78] support vector machine + multiscale permutation entropy Xu et al [79] compound multiscale permutation entropy + particle swarm optimization-support vector machine Li et al [80] improved multiscale permutation + least squares support vector machine Huo et al [81] permutation entropy + Laplacian score + support vector machine Li et al [82] permutation entropy + improved support vector machine Dong et al [83] time-shift multi-scale weighted permutation entropy + gray wolf optimized support vector machine Zhou et al [84] weighted permutation entropy + improved support vector machine ensemble classifier Tiwari et al [85] adaptive neuro fuzzy classifier + multiscale permutation entropy Yi et al [86] tensor-based singular spectrum algorithm + permutation entropy Zhang et al [87] feature space reconstruction + multiscale permutation entropy Zheng et al [88] multi-scale weighted permutation entropy + extreme learning machine Xue et al [89] two-step scheme based on permutation entropy + random forest One typical method is that the wavelet packet transform and decomposition are combined with permutation entropy to enhance the ability of feature extraction [71,72]. The method based on wavelet analysis is effective to extract features contained in the weak transient signal.…”

“…A method based on improved multiscale permutation entropy, laplacian score, and least squares support vector machine-quantum behaved particle swarm optimization is proposed in [80]. Similar to this idea, Huo et al [81] propose a method by integrating the fine-to-coarse multiscale permutation entropy, laplacian score and support vector machine. Multiscale permutation entropy is combined with improved support vector machine based on binary tree for bearing vibration feature extraction, as given in [82].…”