科研之友
微信
新浪微博
Facebook
分享链接
摘要
The paper presents an approach to automatically select the optimal parameters of the kernel functions used in the Support Vector Machines (SVM) approach for the diagnostic task. Various variants of the simulated annealing were implemented and verified in order to obtain the best diagnostic outcomes. The tested system was the fourth order lowpass filter, consisting of two Sallen-Key sections and nine diagnosable elements. The tests covered verification of simulated annealing parameters (starting temperature and annealing ratio) and various SVM kernels (with coding schemes) in the multiple faults detection and location task. The proposed method verified against the exhaustive search.
- 出版日期2011
