Aviation arc fault diagnosis based on weight direct determined neural network

A diagnosis method for the aviation fault arc is proposed in this paper, which is a multi-input weights direct determination (WDD) network. Characteristics of arc current are used for aviation arc fault detection. Arc fault samples are acquired with the help of a self-made arc generator. Feature vectors are obtained from the current of the samples with the wavelet transforming. After training, the fault diagnosis network is verified with the test samples. The result shows that the method has a better performance for aviation fault arc recognition with a simple algorithm.

[1]  Guo Lidong.  Analysis on Time Frequency Feature of Series Arc Fault , 2010 .

[2]  Yunong Zhang,et al.  Weights and structure determination (WASD) of multiple-input hermit orthogonal polynomials neural network (MIHOPNN) , 2012, 2012 24th Chinese Control and Decision Conference (CCDC).

[3]  Yuan-Chun Li,et al.  Arc fault detection based on wavelet packet , 2005, 2005 International Conference on Machine Learning and Cybernetics.

[4]  Liu Guilian,et al.  The Problems and Its Analysis of BP Neural Network , 2010, 2010 International Conference on Intelligent System Design and Engineering Application.

[5]  Zheng Zhi-cheng.  Arc Fault Diagnosis Technology Based on the Analysis of Energy Variation of Signal's Characteristic Frequency Band with Wavelet Transform , 2010 .

[6]  R. Ottoboni,et al.  A novel algorithm for the parallel arc fault identification in DC aircraft power plants , 2012, 2012 IEEE International Instrumentation and Measurement Technology Conference Proceedings.

[7]  Wei Li,et al.  A weights-directly-determined simple neural network for nonlinear system identification , 2008, 2008 IEEE International Conference on Fuzzy Systems (IEEE World Congress on Computational Intelligence).

[8]  Guo Dongsheng.  Weights and structure determination method of multiple-input Sigmoid activation function neural network , 2012 .