|
APPLICATION OF FUZZY NEURAL NETWORKS FOR INSTRUMENT FAULT DIAGNOSIS OF CONDENSATION TURBINE CONTROL
Mariusz PAWLAK*, Jan M. KOŚCIELNY†, Michaeł Z. BARTYŚ†
* Institute of Heat Engineering, Dąbrowskiego 113, 93-208 Łódź, Poland
† Warsaw University of Technology, Faculty of Mechatronics, Institute of Automatic
Control and Robotics, Chodkiewicza 8, 02-525 Warszawa, Poland,
{jmk@mchtr.pw.edu.pl, bartys@mchtr.pw.edu.pl}
In the paper an application of fuzzy neural networks (FNN) for sensor fault diagnosis in condensation turbine control unit was given. The FNN are applied for fault detection and isolation processes. This approach gives the homogenous solution of fault detection and isolation process (FDI). The FNN models of turbine power, live steam pressure and steam mass flow rate were created and verified. Satisfactory models performance indexes were obtained. The fault sensitivity of residuals was investigated and approved.
Keywords: fault diagnosis, power control, power generation, turbines, fuzzy modelling, neural networks, fault tolerant systems
Session slot T-We-A10: Neuro-Fuzzy Applications of Fault Diagnosis/Area code 7e : Fault Detection, Supervision and Safety of Technical Processes
 |
