Application of Variable Precision Rough Set Model and Neural Network to Rotating Machinery Fault Diagnosis

• Published in: Lecture notes in computer science pp. 575 - 584
• Main Authors: Zhou, Qingmin, Yin, Chenbo, Li, Yongsheng
• Format: Book Chapter Conference Proceeding
• Language: English
• Published: Berlin, Heidelberg Springer Berlin Heidelberg 2005; Springer
• Series: Lecture Notes in Computer Science
• Subjects: Algorithmics. Computability. Computer arithmetics; Applied sciences; Computer science; control theory; systems; Decision Table; Exact sciences and technology; Fault Diagnosis; Fault Feature; Precision Level; Theoretical computing; Variable Precision; Rotating machine; Information system; Redundancy; Neural network; Modeling; Rotor; Rough set theory; Dimension reduction; System reduction; Classification; Imperfect information; Shaft; Incomplete information; Fault diagnostic; Power spectrum
• ISBN: 9783540286608; 3540286608; 3540286535; 9783540286530
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/11548706_61

An integration method of variable precision rough set and neural network for fault diagnosis is presented and used in rotary machinery fault diagnosis. The method integrates the ability of variable precision rough set on reduction of diagnosis information system and that of neural network for fault classification. Typical faults of rotating machinery were simulated in our rotor test-bed. The power spectrum data are used as rotating machinery fault diagnosis signal. For inconsistent data and noise data in power spectrum, variable precision rough set model allows a flexible region of lower approximations by precision variables. By attribute reduction based on variable precision rough set, redundant attributes are identified and removed. The reduction results are used as the input of neural network. The diagnosis results show that the proposed approach for input dimension reduction in neural network is very effective and has better learning efficiency and diagnosis accuracy.