Этапы алгоритма квазиоптимального регулирования в системе с импульсным преобразователем

Abstract

Research problem to create a real time quasi-optimal system on the basis of built in microcontrollers is formulated. In the heart of such system lies the identification of a subsystem called ‘the pulse converter – the unknown (including – nonlinear) load’ by the means of an artificial neural network. This subsystem functions under conditions of changing limitations on the impact levels and in the presence of disturbances. The modified algorithm of optimal regulation is suggested. It allows putting the regulation time in its dependence on the size of the initial mismatch without any numerical solving of the transcendental equation and therefore increasing the speed of the system while fulfilling smaller tasks (for example, a task of parameter tracking).