Since thermal plants are multi-input, multi-output systems, it is important
to capture the characteristics of the system for precise combustion contro
l. Furthermore, in refuse incinerator plants (RIPs), the properties of the
fuel are unstable, and minimization of the exhaust emissions is required. T
hus, optimization of efficiency from an overall standpoint, including consi
deration of sensor and control technology, is required in RIPs. In particul
ar, in comparison with stoker incinerators, the combustion cycle in fluidiz
ed bed incinerators (FBIs) is short, and combustion processes occur in mult
iple layers within the incinerator. As a result, analysis of the dynamic ch
aracteristics is considered most the effective way of grasping the characte
ristics of FBIs.
This paper has focused on an operating FBI, where a hybrid system consistin
g of fuzzy systems and neural networks has been realized, which assesses th
e fuel-feeding state on the basis of measured values and combustion image p
rocessing, and operates with low CO/NOx concentrations by means of air-fuel
ratio control. Furthermore, a proposed tuning method for the fuzzy systems
simplifies the evaluation of speculative results, and the determination of
control rules, by the utilization of an operation support system based on
a numerical model. (C) 1998 Elsevier Science Ltd. All rights reserved.