Abstract:

Considering the operation of the marine supercharged boiler in a wide range of conditions, a multi-model set was built for the drum water level control using the combination of the fixed models for the typical operation conditions and the radial basis function neural network(RBFNN). The fixed models can not adapt to the wide range changes of the operating conditions, so using the RBFNN to compensate the dynamic prediction errors. A multi-model drum water level controller was developed using the predictive control algorithm in the state space. The simulation results show that the developed controller can control stably the drum water level, improve the system response, solve the control difficulty under wide-range variable operating conditions.

Key words: automatic control technology, boiler, drum water level, predictive control, multi-model control