Kourounakis, Nicolaos P.2024-08-142024-08-1419981998https://hdl.handle.net/1828/18489This thesis presents a novel model-based fault detection system specifically designed to offer improved fault detection for the reverse pilot of cable television networks. The approach, however, is general and can be applied to a variety of large scale engineering plants. For the modeling part of the system, feedforward neural networks and statistical analysis techniques are employed. A description of these techniques as well as the fault detection methodology used is provided. Finally, this thesis presents the results of a large scale experiment on which the fault detection properties of the developed technique were evaluated and compared to those of limit checking fault detection.128 pagesAvailable to the World Wide WebThesis