A neuro-fuzzy controller for a stoker-fired boiler, based on behavior modeling

Abstract

A key issue in an industrial stoker-fired boiler is the design of an efficient and robust controller for its combustion system, so that the boiler can provide a continuous supply of steam at the desired pressure conditions. However, it is difficult to achieve this objective by using a model-based approach because of the high nonlinearity and uncertainty of boiler systems. In addition, the control performance may also suffer as a result of strong load changes, large disturbances, large time lags, and so forth. This paper presents a behavior-modeling-based approach to the design of a neuro-fuzzy controller for the combustion control of a stoker-fired boiler. In this approach, boiler combustion processes with unknown structure are modeled by defining three dynamic behaviors. According to these behavior ‘templates’, their corresponding fuzzy-logic controllers can be optimized off-line. During boiler system operation, the appropriate fuzzy-logic controller is fired, based on an on-line assessment of its dynamic behavior. The application results obtained demonstrate the effectiveness and the robustness of the proposed controller.