| Summary: | 博士 === 國防大學理工學院 === 國防科學研究所 === 99 === This thesis applied Support Vector Machines (SVM) and Support Vector Regression (SVR) to the fault diagnosis, the adaptive intelligent control, and the Inverse Heat Conduction (IHCP) problems in the military areas. For fault diagnosis, a case study of a field air defense gun was provided. The maintenance history data was used to construct an efficient fault diagnosis system by SVM which can discovers the fault part precisely and rapidly by input the measurement data. The experimental results show that SVM models diagnose faults more accuracy than BPN models when numbers of training sets are small. For adaptive intelligent control problem, this thesis applies SVR technique to design an on-line learning adaptive intelligent controller, called SVR controller (SVRC) by on-line learning to construct the adaptive adjustable rules to perform a DC motor speed control. The simulation shows superior control efficiency. In the IHCP, this thesis applied SVR algorithm to estimate the unknown time dependent heat flux boundary conditions of one dimensional and two dimensional IHCP. This research first observed the inverse relation of temperature variation and heat flux. Construct the inverse estimating model by SVR using prepared temperature/heat flux data for training relations then estimate heat flux by input the measured temperature. From simulations results show, the SVR algorithm solving IHCP has great advantages not only convenient to use, accurate calculation, but the ability of regularization to overcome the ill-posed of IHCP.
|
|---|
