Investigation of a hybrid switching control system

A servo motor is to be used to position the cutting arm in a hypothetical pattern generation application. The motor is controlled in closed-loop in order to track, with zero asymptotic error, a reference signal represented by either a sinusoidal, triangular, or square wave. In addition, the schedule of reference signal type changes is not known a priori and the controlled system must achieve asymptotic tracking without operator intervention. As no simple single controller can satisfy these requirements for all setpoint types, a Hybrid Switching Control System is proposed which combines intuitive logic with standard control techniques. Under the guidance of a simple supervisor, the controller corresponding to each type of setpoint is switched in and out of the active feedback loop as required. A simple Multi-layer Perceptron neural network was selected to identify the type of signal being tracked and hence initiate controller switching. This network performed very well even in the presence of measurement noise, and the hybrid system automatically tracked each of the three types of reference signal over a wide range of signal amplitude and frequency. However, the reconfiguration interval was quite long (although still acceptable in terms of the proposed application), and the size of the neural net structure had to be limited for the system to work in real-time.

Grantor dc:publisher.institution

Department of Electrical Engineering

Year dc:date.issued

1996

Author dc:creator

• Narożny, Jacek Andrzej

Advisor dc:contributor.advisor

• Braae, Martin

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/11427/20448

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/20448