online read us now
Paper details
Number 1 - June 1992
Volume 2 - 1992
Fault detection in coal fired power plants using nonlinear filtering
Zohreh Fathi, Józef Korbicz, W. Fred Ramirez
Abstract
On the contrary to many recent attempts using knowledge-based system techniques where diagnostic analysis is based solely on measurable and observable data, in this work we propose to investigate the adaptive inclusion of a state and/or parameter estimation module in the diagnostic reasoning loop, in addition to employing information based on measurable data. The design methodology is a new layered knowledge base that houses heuristics knowledge in the high-levels and the process-general estimation knowledge
in the low-levels. The purpose of this paper is to present the failure detection issues of the deaerator control subsystem for the coal fired power plant. The main emphasis is placed upon the model-based redundancy methods which create the low-levels of the knowledge base. Due to the highly nonlinear nature of the power plant dynamic, the modified extended Kalman filters are designed for use as detection filters. The developed approach is shown to be effective in detecting and isolating failures of a subsystem of a power plant with an appropriate degree of complexity.
Keywords
-