Kernel Principal Component Analysis-Based Method for Fault Diagnosis of SINS
Fault diagnosis is necessary in inertial navigation system to ensure navigation success. The kernel principal component analysis (KPCA) method is applied to fault diagnosis of inertial navigation system. At first, the square prediction error is used as the fault monitoring indictor. In addition, this paper locates the faults by using sensor variables changes. In order to reduce the dependence to the prior knowledge in selecting kernel function parameter, genetic algorithm is introduced into the Gauss RBF kernel function parameter optimization. It can improve the scientific nature of parameter choice. Experimental results show that the method based on KPCA of the inertial navigation system has good fault monitoring and recognition ability.
KeywordsStrap down inertial navigation system Sensor Fault diagnosis Kernel principal component analysis Genetic algorithm
- 2.Gao D, Wu C-g, Zhang B-k (2011) Method combining PCA and SDG for fault diagnosis of sensors and its application. J Syst Simul 23(3):567–573Google Scholar
- 5.Gao Y-g, Li Z Shi D-q (2008) A hybrid learning method for RBFNN based on AGA. Radio Eng China 38(11):25–28Google Scholar
- 6.Deng X-g, Tian X-m (2005) Nonlinear process fault diagnosis method using kernel principal component analysis. J Shandong Univ (Eng Sci) 35(3):103–106Google Scholar