Nonlinear control system with neural network controller using RasVal learning

In this paper, a new learning algorithm is applied to the design of the optimal neural controller of a nonlinear control system. The optimization method is called RasVal; it is a kind of random search process for a global minimum in a single framework. The searching for a global minimum based on the probability density functions can be modified using information on the success or failure of the past searching in order to execute an intensified and diversified search. By applying the proposed method to a nonlinear crane control system which can be controlled by the universal learning network with sigmoid functions, it is shown that RasVal is superior in performance to the commonly used backpropagation learning algorithm