scholarly journals NC Machine Tools Fault Diagnosis Based on Kernel PCA andk-Nearest Neighbor Using Vibration Signals

2015 ◽  
Vol 2015 ◽  
pp. 1-10 ◽  
Author(s):  
Zhou Yuqing ◽  
Sun Bingtao ◽  
Li Fengping ◽  
Song Wenlei
Keyword(s):  
Fault Diagnosis ◽  
Machine Tools ◽  
Nearest Neighbor ◽  
Principal Component ◽  
Feature Space ◽  
Kernel Principal Component Analysis ◽  
Dimensional Manifold ◽  
Nc Milling ◽  
Nc Machine ◽  
Nc Machine Tools

This paper focuses on the fault diagnosis for NC machine tools and puts forward a fault diagnosis method based on kernel principal component analysis (KPCA) andk-nearest neighbor (kNN). A data-dependent KPCA based on covariance matrix of sample data is designed to overcome the subjectivity in parameter selection of kernel function and is used to transform original high-dimensional data into low-dimensional manifold feature space with the intrinsic dimensionality. ThekNN method is modified to adapt the fault diagnosis of tools that can determine thresholds of multifault classes and is applied to detect potential faults. An experimental analysis in NC milling machine tools is developed; the testing result shows that the proposed method is outperforming compared to the other two methods in tool fault diagnosis.

Fault Diagnosis of Bearing Based on KPCA and KNN Method

2014 ◽  
Vol 986-987 ◽  
pp. 1491-1496 ◽  
Author(s):  
Qiang Wang ◽  
Yong Bao Liu ◽  
Xing He ◽  
Shu Yong Liu ◽  
Jian Hua Liu
Keyword(s):  
Principal Component Analysis ◽  
Fault Diagnosis ◽  
Nearest Neighbor ◽  
Feature Vector ◽  
Principal Component ◽  
Feature Space ◽  
Component Analysis ◽  
Kernel Functions ◽  
Kernel Principal Component Analysis ◽  
K Nearest Neighbor

Selection of secondary variables is an effective way to reduce redundant information and to improve efficiency in nonlinear system modeling. The combination of Kernel Principal Component Analysis (KPCA) and K-Nearest Neighbor (KNN) is applied to fault diagnosis of bearing. In this approach, the integral operator kernel functions is used to realize the nonlinear map from the raw feature space of vibration signals to high dimensional feature space, and structure and statistics in the feature space to extract the feature vector from the fault signal with the principal component analytic method. Assessment method using the feature vector of the Kernel Principal Component Analysis, and then enter the sensitive features to K-Nearest Neighbor classification. The experimental results indicated that this method has good accuracy.

scholarly journals Modified Kernel Marginal Fisher Analysis for Feature Extraction and Its Application to Bearing Fault Diagnosis

2016 ◽  
Vol 2016 ◽  
pp. 1-16 ◽  
Author(s):  
Li Jiang ◽  
Shunsheng Guo
Keyword(s):  
Feature Extraction ◽  
Fault Diagnosis ◽  
Nearest Neighbor ◽  
Feature Space ◽  
High Dimensional ◽  
Dimensional Manifold ◽  
Bearing Fault ◽  
Bearing Fault Diagnosis ◽  
Low Dimensional ◽  
Low Dimensional Manifold

The high-dimensional features of defective bearings usually include redundant and irrelevant information, which will degrade the diagnosis performance. Thus, it is critical to extract the sensitive low-dimensional characteristics for improving diagnosis performance. This paper proposes modified kernel marginal Fisher analysis (MKMFA) for feature extraction with dimensionality reduction. Due to its outstanding performance in enhancing the intraclass compactness and interclass dispersibility, MKMFA is capable of effectively extracting the sensitive low-dimensional manifold characteristics beneficial to subsequent pattern classification even for few training samples. A MKMFA- based fault diagnosis model is presented and applied to identify different bearing faults. It firstly utilizes MKMFA to directly extract the low-dimensional manifold characteristics from the raw time-series signal samples in high-dimensional ambient space. Subsequently, the sensitive low-dimensional characteristics in feature space are inputted into K-nearest neighbor classifier so as to distinguish various fault patterns. The four-fault-type and ten-fault-severity bearing fault diagnosis experiment results show the feasibility and superiority of the proposed scheme in comparison with the other five methods.

Research on Fault Diagnosis of High-Voltage Circuit Breaker in Electrical Power System Based on KPCA

2013 ◽  
Vol 676 ◽  
pp. 269-272
Author(s):  
Yi Ping Wang ◽  
Fei Mei ◽  
Jian Yong Zheng ◽  
Jun Mei
Keyword(s):  
Fault Diagnosis ◽  
Power Systems ◽  
High Voltage ◽  
Electrical Power ◽  
Circuit Breaker ◽  
Principal Component ◽  
Feature Space ◽  
Original Data ◽  
Kernel Principal Component Analysis ◽  
Electrical Power System

High-voltage circuit breakers (HVCBs) play an essential role in electrical power systems, which can ensure and control the smooth operation of power grids. Therefore, a fault diagnosis method of HVCBs based on Kernel Principal Component Analysis (KPCA) is proposed in this paper. As the fault data of HVCBs have the characteristic of multi-dimensional nonlinearity, the proposed method calculates inner kernel function of HVCBs' closing current in the original data space so as to achieve nonlinear mapping to the feature space of the original data. Afterwards, feature extraction and pattern classification of fault data can be accomplished in the feature space by means of monitoring SPE statistics. Experiment results have proved that the KPCA method can effectively improve the precision of fault diagnosis of HVCBs.

Two-parameter KNN algorithm and its application in recognition of brand rice

2021 ◽  
pp. 1-7
Author(s):  
Zhu Siyu ◽  
He Chongnan ◽  
Song Mingjuan ◽  
Li Linna
Keyword(s):  
Near Infrared ◽  
Nearest Neighbor ◽  
Recognition Accuracy ◽  
Principal Component ◽  
Kernel Principal Component Analysis ◽  
K Nearest Neighbor ◽  
Fisher Discriminant Analysis ◽  
Pattern Class ◽  
Fisher Discriminant ◽  
Two Parameter

In response to the frequent counterfeiting of Wuchang rice in the market, an effective method to identify brand rice is proposed. Taking the near-infrared spectroscopy data of a total of 373 grains of rice from the four origins (Wuchang, Shangzhi, Yanshou, and Fangzheng) as the observations, kernel principal component analysis(KPCA) was employed to reduce the dimensionality, and Fisher discriminant analysis(FDA) and k-nearest neighbor algorithm (KNN) were used to identify brand rice respectively. The effects of the two recognition methods are very good, and that of KNN is relatively better. Howerver the shortcomings of KNN are obvious. For instance, it has only one test dimension and its test of samples is not delicate enough. In order to further improve the recognition accuracy, fuzzy k-nearest neighbor set is defined and fuzzy probability theory is employed to get a new recognition method –Two-Parameter KNN discrimination method. Compared with KNN algorithm, this method increases the examination dimension. It not only examines the proportion of the number of samples in each pattern class in the k-nearest neighbor set, but also examines the degree of similarity between the center of each pattern class and the sample to be identified. Therefore, the recognition process is more delicate and the recognition accuracy is higher. In the identification of brand rice, the discriminant accuracy of Two-Parameter KNN algorithm is significantly higher than that of FDA and that of KNN algorithm.

Sign in / Sign up

Export Citation Format

Share Document