Синтез нейросетевых подходов управления сложными динамическими процессами в сахарном производстве

Abstract

В работе были рассмотрены вопросы управления сложными динамическими процессами, имеющими место в сахарном производстве. Сделан анализ существующих методов управления, а также сделан анализ управления процессами в сахарном производстве на основе применяемых нейронных сетей. Детально проанализированы различные подходы в системе нейроуправления технологическими процессами в сахарном производстве. Полученные алгоритмы, в соответствии с критериями качества управления, позволяют учитывать изменение входных и выходных сигналов в системе управления процессами.

The question of difficult and nonstationary dynamical processes that take place in sugar production have been examined in this work. There have been determined basic, commonly used, effective methods and schemes of neurocontrol of technological processes: direct and indirect methods, consecutive scheme of neurocontrol, scheme of inverse spread in time, parallel neurocontrol and direct accommodative control. For each of examined methods there have been handpicked neural network models. There have been made the analysis of these models, also the analysis of controlling of process in sugar production based on applied neural networks. There have been particularly analyzed different approaches in neurocontrol of technological processes. There have been received algorithms of settings of neuroregulators according to appropriate quality criteria of controlling, also allows to take into consideration changes of incoming and outgoing signals in automatic system of controlling of processes. Improvement of properties of obtained algorithms reaches by using methods of controlling theory, by means of taking into consideration not only the incoming and outgoing values, but also their foretold values.