Zhiliang Liu

Feature selection for fault level diagnosis of planetary gearboxes

Feature selection is critical to maintain high performance of classification-based fault diagnosis with a large feature size. In this paper, we propose a criterion to evaluate features effectiveness by class separability that is defined on cosine similarity in the kernel space of the Gaussian radial basis function. We develop a feature selection algorithm accordingly using the proposed criterion together with sequential backward selection and a feature re-ranking mechanism. We then employ the proposed feature selection algorithm to determine fault-sensitive features and select them for fault level diagnosis of planetary gearboxes. The experimental results demonstrate that the proposed algorithm can effectively reduce the feature size and improve accuracy of fault level diagnosis simultaneously.

Parameter selection for Gaussian radial basis function in support vector machine classification

The Gaussian radial basis function is widely used in the support vector machine (SVM) due to its attractive characteristics. The parameter (σ) in this kernel is crucial to robust performance of SVM. In this paper, we derive a formula to compute the optimal s under the principle of maximizing the class separability in the kernel space. The most attractive feature of the proposed method is that no optimization search algorithm is required in parameter selection; and thus our method is computational effective. The experimental results demonstrate the proposed method is fast and robust.

Remaining useful life prediction of rolling element bearings based on health state assessment

Instead of looking for an overall regression model for remaining useful life (RUL) prediction, this paper proposes a RUL prediction framework based on multiple health state assessment that divides the entire bearing life into several health states where a local regression model can be built individually. A hybrid approach consisting of both unsupervised learning and supervised learning is proposed to automatically estimate the real-time health state of a bearing in cases with no prior knowledge available. Support vector machine is the main technology adopted to implement health state assessment and RUL prediction. Experimental results on accelerated degradation tests of rolling element bearings demonstrate the effectiveness of the proposed framework.

Feature ranking for support vector machine classification and its application to machinery fault diagnosis

This article provides a feature ranking criterion for multi-class support vector machine classification. In the proposed criterion, feature effectiveness is estimated for individual features by their contributions to class separability in the kernel space. Class separability, measured by cosine similarity, is defined by an objective function that consists of within-class and between-class separabilities. Feature ranking is achieved for individual features by sorting their effectiveness scores. The proposed criterion is validated on University of California Irvine benchmark datasets and also applied to pitting diagnosis for a planetary gearbox. The experimental results demonstrate that the proposed criterion of feature ranking is computationally economic and effective.

Fault diagnosis for planetary gearboxes using multi-criterion fusion feature selection framework

Feature selection has been used to achieve dimension reduction in the field of fault diagnosis. This article introduces a multi-criterion fusion framework for feature selection that takes into account three aspects of features: effectiveness, correlation, and classification performance. This framework enables a more comprehensive evaluation of features than does a single criterion. The proposed framework is implemented using five effectiveness criteria and a correlation criterion. It is used to diagnose eight failure modes of a planetary gearbox. The experimental results demonstrate that the proposed multi-criterion framework outperforms many well-studied single criteria.

A Gaussian radial basis function based feature selection algorithm

Recently Li et al. proposed a parameter selection method for Gaussian radial basis function (GRBF) in support vector machine (SVM). In his paper cosine similarity was calculated between two vectors based on the properties of GRBF kernel function. Lis method can determine an optimal sigma in SVM and thus efficiently improve its performance, yet it is limited by only focusing on a fixed original feature space and may suffer if the space contains some irrelevant and redundant features, especially in a high-dimensional feature space. In this paper, Lis method is extended to a flexible feature space so that feature selection and parameter selection are conducted at the same time. A feature subset and sigma are determined by minimizing the objective function that considers both within-class and between-class cosine similarities. Our experimental results demonstrate that the proposed method has a better performance than Lis method and traditional SVM in terms of classification accuracy.

Diagnosis of pitting damage levels of planet gears based on ordinal ranking

Information on damage levels is useful for condition based preventive maintenance decision making. To diagnose the damage level of machinery, classification methods have been widely employed. However, classification methods couldnt utilize the ordinal information contained in damage levels. Ordinal ranking, a recently studied supervised learning method, utilizes the ordinal information in the data, which makes it useful in diagnosing damage levels. This paper applied a reported ordinal ranking algorithm to diagnose the pitting damage levels in a planet gear for the first time. Experiment results show that ordinal ranking can generate a good ranking model for damage level diagnosis. Comparisons with a classical classification method demonstrate the advantage of ordinal ranking.

An Analytical Approach to Fast Parameter Selection of Gaussian RBF Kernel for Support Vector Machine

The Gaussian radial basis function (RBF) is a widely used kernel function in support vector machine (SVM). The kernel parameter σ is crucial to maintain high performance of the Gaussian SVM. Most previous studies on this topic are based on optimization search algorithms that result in large computation load. In this paper, we propose an analytical algorithm to determine the optimal σ with the principle of maximizing between-class separability and minimizing within-class separability. An attractive advantage of the proposed algorithm is that no optimization search process is required, and thus the selection process is less complex and more computationally efficient. Experimental results on seventeen real-world datasets demonstrate that the proposed algorithm is fast and robust when using it for the Gaussian SVM.

Notice of Retraction A study on applications of principal component analysis and kernel principal component analysis for gearbox fault diagnosis

Principal component analysis (PCA) and kernel principal component analysis (KPCA) are widely used approaches of dimensionality reduction. They have been demonstrated useful for gearbox fault diagnosis. This paper provides a brief review of applications of PCA and KPCA for gearbox fault diagnosis. Literature is mainly grouped into two categories: applications of the conventional PCA/KPCA and applications of the improved PCA/KPCA. Discussions about the future work of PCA/KPCA on gearbox fault diagnosis are also provided in this paper.

Classification of gear damage levels in planetary gearboxes

Linear discriminant analysis (LDA) is a method of feature extraction that has demonstrated successful applications. The selection of the number of discriminant directions (r) is important to LDA, yet little attention is paid in the reported literature. In this paper a method is proposed for determining the optimal r in terms of the classification accuracy of support vector machine. The method is applied to identify gear damage levels in a planetary gearbox. Planet gears with four damage levels labeled as baseline, slight, moderate, and severe were used in lab experiments for data collection. Results demonstrate that the proposed method outperforms two reported methods and is effective to address the given problem.