A fault diagnosis model based on singular value manifold features, optimized SVMs and multi-sensor information fusion

Abstract

To achieve better fault diagnosis of rotating machinery, this paper presents a novel intelligent fault diagnosis model based on singular value manifold features (SVMF), optimized support vector machine (SVMs) and multi-sensor information fusion. Firstly, a new fault feature denoted as SVMF is developed to better represent faults. SVMF is acquired by extracting manifold topology features of the singular spectrum. Compared with frequently-used fault features, the feature scale of SVMF is constant under variable rotating speed, and the extraction process of SVMF also has the effect of self-weighting. So SVMF has a better representation of faults. Then, to select optimal parameters for model training of SVMs, an improved fruit fly algorithm is proposed by introducing guidance search mechanism and enhanced local search operation, and as a result, both the convergence speed and accuracy are improved. At last, the Dempster-Shafer evidence theory is introduced to fuse decision-level information of SVMs models of multiple sensors. By information fusion, the conflict of fault diagnosis conclusions from multi-sensor is eliminated, which lead to high robustness and accuracy of fault diagnosis model. As a summary, the proposed method combines the advantages of SVMF in fault representation, SVMs in fault identification and the Dempster-Shafer evidence theory in information fusion, and as a result the proposed method will get better fault diagnosis performance. The proposed intelligent fault diagnosis model is subsequently applied to the fault diagnosis of the gearbox. Experimental results show that the proposed fault diagnostic framework is versatile at detecting faults accurately.

A sensor fusion based approach for bearing fault diagnosis of rotating machine

“…In Azamfar et al, 30 CNN-Softmax based diagnostic method is used only current signature. In Su et al 42 DS-SVM, the researchers used only vibration data. As per the available literature, it is challenging to fuse vibration and thermal image data using these techniques in their indigenous form.…”

“…41 Dempster–Shafer’s (D-S) evidence theory is a remarkable approach for decision level fusion to deal with ambiguous information. 42 However, classical D-S evidence theory may lead to incomprehensible results when using information from different sensors. 43 An improved approach with classical D-S evidence theory using Euclidian distance to distinguish various parts of evidence was used to cope with uncertainty in the decision level fusion.…”

Multi-channel data fusion and intelligent fault diagnosis based on deep learning

“…In the setting of fault diagnosis modeling, the extracted features are used to train the classifier. The commonly used classification models include support vector machine (SVM), K-nearest neighbor and nearest neighbor (NN) [6][7][8]. However, the multi-channel data cannot be processed directly by these common traditional methods due to the complex tensor structure.…”

Measuring the Uncertainty in the Original and Negation of Evidence Using Belief Entropy for Conflict Data Fusion

“…The Dempster-Shafer evidence theory was first proposed by Dempster [ 13 ] in 1967, and then further developed by his student Shafer [ 14 ] in 1976. This theory is a reasoning theory that can effectively deal with uncertain information [ 20 , 21 ], and is widely used in many fields, such as fault diagnosis [ 22 , 23 ], decision making [ 24 , 25 , 26 ], risk assessment [ 27 , 28 ], classification [ 29 , 30 , 31 ], and so on [ 32 , 33 ], which solves many problems caused by uncertain information. However, in application, the classic combination rule of DS theory has been found to have some problems.…”