Application of improved RBFNN in comprehensive evaluation for maintenance quality

Author

Wang, Shengfeng ; Wang, Hongwei ; Ni, Mingfang ; Kou, Deqi ; Tong, Xun

Author_Institution

Dept. of Tech. Support Eng., Acad. of Armored Forces Eng., Beijing, China

fYear

2011

fDate

17-19 June 2011

Firstpage

770

Lastpage

773

Abstract

According to the characteristics of evaluation of maintenance quality, in this paper partial least squares (PLS) is adopted to improve the common least squares (LS), and the maintenance quality evaluation model based on FCM-PLS-RBFNN is set up, and the learning and training algorithm is provided for FCM-PLS-RBFNN, and the improving effect of the model and its validity and precision in maintenance quality evaluation is tested by the living example of certain equipment maintenance quality comprehensive evaluation. The result shows that the FCM-PLS-RBFNN is faster than FCM-LS-RBFNN in learning, and its approaching ability and popularize performance are improved obviously. It is workable and effective to apply the FCM-PLS-RBFNN in modeling and evaluating for maintenance quality. It provides new ideas for researching on the more external and better popularizes maintenance quality evaluation method.

Keywords

fuzzy set theory; learning (artificial intelligence); maintenance engineering; pattern clustering; radial basis function networks; RBFNN; equipment maintenance quality comprehensive evaluation; fuzzy c-mean clustering; learning; maintenance quality evaluation model; partial least squares; training algorithm; Data models; Indexes; Maintenance engineering; Mathematical model; Neural networks; Testing; Training; RBF neural networks; comprehensive evaluation; fuzzy c-means clustering; maintenance quality; partial least squares;

fLanguage

English

Publisher

ieee

Conference_Titel

Quality, Reliability, Risk, Maintenance, and Safety Engineering (ICQR2MSE), 2011 International Conference on

Conference_Location

Xi´an

Print_ISBN

978-1-4577-1229-6

Type

conf

DOI

10.1109/ICQR2MSE.2011.5976723

Filename

5976723