MKL and OC-ELM fault detection based on lp-norm constraint

被引：0
|
作者：
Liu X. [1 ]
Zhao J.-Y. [1 ]
Zhu M. [2 ]
Zhang W. [3 ]
机构：
[1] Naval Aviation University, Yantai
[2] Unit 91576 of the PLA Troops, Ningbo
[3] Military Representative Office of Naval Equipment Department in Xianyang, Xianyang
来源：
Liu, Xing (xinghandeqipan@sina.com) | 1600年 / Northeast University卷 / 36期
关键词：
Extreme learning machine; Fault detection; l[!sub]p[!/sub]-norm constraint; Multiple kernel learning; One-class classification;
D O I：
10.13195/j.kzyjc.2020.0443
中图分类号：
学科分类号：
摘要：
Aiming at the problems of the shortage of fault samples for active new equipment and the low accuracy of existing algorithms for fault detection, the multiple kernel learning (MKL) and the one-class extreme learning machine (OC-ELM) are combined, and the lp-norm constrainted multiple kernel learning one-class ELM (lp-MKOCELM) is proposed. Under the lp-norm constraint, a mathematical optimization form combining the MKL and the OC-ELM is defined, and the update method of combination weights of the base kernel and Lagrange multipliers are derived. To facilitate the implementation of fault detection, the test statistic and detection threshold based on the lp-MKOCELM are defined. The approximate equivalence of different norm constraints is confirmed through experiments. The proposed method is applied to the commonly used UCI data set and test data of an equipment. The experimental results show that, compared with the traditional SVDD, PCA, OC-SVM, and OC-KELM, the proposed method can significantly improve the detection accuracy while balancing missing alarm and false alarm. © 2021, Editorial Office of Control and Decision. All right reserved.
引用
收藏
页码：2379 / 2388
页数：9
相关论文
共 21 条
  • [1] Pimentel M A F, Clifton D A, Clifton L, Et al., A review of novelty detection, Signal Processing, 99, pp. 215-249, (2014)
  • [2] Bing Q C, Gong B W, Yang Z S, Et al., A short-term traffic flow local prediction method of combined kernel function relevance vector machine, Journal of Harbin Institute of Technology, 49, 3, pp. 144-149, (2017)
  • [3] Li J, Li D C., Wind power time series prediction using optimized kernel extreme learning machine method, Acta Physica Sinica, 65, 13, pp. 39-48, (2016)
  • [4] Moya M, Koch M, Hostetler L., One-class classifier networks for target recognition applications, Proceedings of the World Congress on Neural Networks, pp. 797-801, (1993)
  • [5] Tax D M J, Duin R P W., Support vector data description, Machine Learning, 54, 1, pp. 45-66, (2004)
  • [6] Navarro-Acosta J A, Resendiz-Flores E O., Gompertz binary particle swarm optimization and support vector data description system for fault detection and feature selection applied in automotive pedals components, The International Journal of Advanced Manufacturing Technology, 92, 5, pp. 2311-2324, (2017)
  • [7] Scholkopf B, Platt J C, Shawe-Taylor J, Et al., Estimating the support of a high-dimensional distribution, Neural Computation, 13, 7, pp. 1443-1471, (2001)
  • [8] Dai H F, Bian H W, Ma H, Et al., Information fault detection for integrated navigation systems using one-class support vector machine, Journal of Chinese Inertial Technology, 25, 4, pp. 555-560, (2017)
  • [9] Zhu W, Zhong P., A new one-class SVM based on hidden information, Knowledge-Based Systems, 60, pp. 35-43, (2014)
  • [10] Mefraz Khan N, Ksantini R, Shafiq Ahmad I, Et al., linebreak Covariance-guided one-class support vector machine, Pattern Recognition, 47, 6, pp. 2165-2177, (2014)
123