Implementation of a distributed control system using real-time operating system

Official URL: http://risc01.sabanciuniv.edu/record=b1169197 (Table of Contents)

In a typical distributed control system, computer nodes communicate through a common communication channel that introduces data loss and random delays. Supplying a generic solution to these constraints is hard due to the complexity and large variety of possibilities that may affect these constraints in real life applications. In a modern communication network, if data is corrupted during transmission, it can be resent. However, it is not feasible to retransmit in control applications; if the packet contents correspond to measured plant outputs, then the most recent data should be measured and sent instead, or if the packet contents correspond to a control signal and the retransmission would cause the control signal to be applied late to the plant, it would be better to recalculate the signal and send it again. This thesis is an attempt to implement a distributed control system design method, Model Based Predictive Networked Control System (MBPNCS), which accepts the fact that arbitrary delay and data loss may happen. The MBPNCS method approaches the problem by using a plant model to predict a predefined number of future states of the plant and respective control signal for each, to compensate for the possible delay and data loss that can take place during the communication between nodes. In this work, after previous works have been examined, predictive control method that is used in the implementation is introduced. Design and implementation of the methodology is explained in detail and results of the tests are presented.