Lyashenko A. The sugar production extraction plant neural control system synthesis based on radial basis function network

The dissertation is devoted to the neural networks approach development for solving the problems of sugar production diffusion station technological processes automated control. The existing management control systems analysis was carried out and on its base the conclusion about the appropriateness of an adaptive control methods developing using the artificial neural networks theory was drawn. Examination of static and dynamic nueral networks properties caused using of zero and first order radial-basis network to solve the assigned task. A basis function piecewise-linear approximation was suggested and that allowed to shorten the nueral network training process and reduce its learning duration. The relations that allow to implement neural network and neural network sub-controllers are obtained. The indirect neural control procedures based on zero and first-order radial-basis network with basis function approximation was suggested. The experimental studies and implementation confirm the results reliability.