透過您的圖書館登入
IP:18.223.209.254
  • 期刊

Fault Diagnosis of Ship Regional Distribution Power System based on Multi Granularity ELM

摘要


In view of the characteristics of multi-type and complex samples in fault diagnosis of marine regional distribution power system, a multi granularity extreme learning machine (MGELM) model is proposed to apply to fault diagnosis. The simulation model of ship ring area distribution system is built by SIMULINK, the voltage and current of distribution line and the energy signal characteristics of generator set are extracted by wavelet packet decomposition; Based on the multi granularity feature subset obtained by neighborhood rough set attribute reduction, the ELM network is improved, and combined with the reliability weighted voter, the MGELM model is constructed to diagnose the ship regional distribution fault. Compared with extreme learning machine, support vector machine and random forest model, the experimental results show that the proposed method improves the model training stability by more than 80% compared with extreme learning machine, it has higher accuracy and reliability compared with all comparison models in classification performance, and can still maintain the optimal classification effect when the samples are unbalanced.

參考文獻


G.B. Huang, Q.Y. Zhu, C.K. Siew. Extreme learning machine: Theory and applications, Neurocomputing, Vol. 70 (2006), p.489-501.
Q.H. Hu,D.R. Yu,J.F. Liu,C.X. Wu. Neighborhood Rough Set based Heterogeneous Feature Subset Selection, Information Sciences, Vol. 178 (2008) No.18, p.3577-3594.
X.Ying. An Overview of Overfitting and its Solutions,Journal of Physics: Conference Series, Vol. 1168(2019) No.2, p.1-6.
D.M. Hawkins. The Problem of Overfitting,Journal of Chemical Information and Computer Sciences, Vol. 44 (2004) No.1, p.1-12.
Y. Dai:Research on High Reliability Regional Power Distribution Technology of Ship Power System (MS., Jiangsu University of Science and Technology, China 2019).

延伸閱讀