Adaptive monitoring of emissions in energy boilers using self-organizing maps: An application to a biomass-fired CFB (circulating fluidized bed)

Abstract

Improvement of energy efficiency, reduction of operating costs, and reduction of harmful emissions released into the atmosphere are issues of major concern in modern energy plants. While air emissions have to be restricted due to tightening environmental legislation, at the same time it is ever more important to be able to respond quickly to any changes in the load demand or fuel quality. As unpredictability increases with changing fuel quality and more complex operational strategies, undesired phenomena such as increased emission release rates may become more likely. Therefore, it is crucial that emission monitoring systems are able to adapt to varying conditions, and advanced methodologies are needed for monitoring and decision-support. In this paper a novel approach for advanced monitoring of emissions in CFB (circulating fluidized bed) boilers is described. In this approach a model based on SOM (self-organizing maps) is updated regularly to respond to the prevailing condition of the boiler. After creating each model a new set of measurements is input to the system, and the current state of the process is determined using vector distance calculation. Finally, the system evaluates the current condition and may alert if a preset limit defined for each emission component is exceeded.