Design of ball and beam system controller based on kernel function principal component analysis

To simulate the actions of human, in this paper a controller of the ball and beam system has been designed based on kernel function principal component analysis (KPCA). The sampling data were acquired by controlling ball and beam system manually, the time sequence of embedded dimensionality were determined by fault nearest neighbours dot algorithm, their features were extracted in the space of nonlinear supporting variable through KPCA algorithm and phase-space are reconstructed, then the linear regression were realized by the least square method and mathematical model was built. Experiments indicated that the controller was effective. It can simulate the behavior of human and possesses high precision and stability.