Implementation of an adaptive intelligent controller for benchmark thermal system

In this work, a neural network (NN) based adaptive controller is developed and implemented for precise temperature control of a benchmark thermal system in cold climate. The newly devised NN controller is capable of overcoming the limitations of model dependent conventional fixed gain temperature controllers. The proposed NN controller is designed using the combination of off-line and on-line trainings of the feed-forward neural network. The transient and steady-state behaviors of the proposed NN-based thermal control system for central heating are improved by incorporating a unique feature of adaptive learning which aids the on-line robust temperature control over a wide operating range. The stability of the proposed NN-based thermal system has been ensured by a combination of off-line and on-line trainings of the NN. As an integral part of this work, efforts have been directed for the real-time implementation of the NN-based thermal system using a digital signal processor (DSP) controller board ds1102. A series of tests have been carried out in order to evaluate the performances of the NN-based benchmark thermal system for central heating. The laboratory test results validate the efficiency of the NN controller as an adaptive controller in the high performance benchmark thermal systems.