Soft-sensor design and dynamic model development for a biomass combustion power plant.
Dissertation, Universität des Saarlandes, Saarbrücken, Germany, Sept. 11st, 2017.
From a system theory perspective, a biomass power plant is a nonlinear,coupled multivariate system with multiple inputs (fuel feed, air supply, grate speed) and multiple outputs (gas temperature, oxygen concentration, steam generated), where the different process input-output relationships are difficult to understand due to the large disturbances acting on the combustion process, which emanate mainly from the varying calorific value of fuel delivered to the furnace. Hence, any attempt to maintain stable operating conditions and to design or improve the control strategy being employed will lead to suboptimal solutions, which may jeopardize the commercial character of the combustion site. One possible way to handle such a situation is by improving the combustion performance using advanced model-based control strategies for this aim to further ameliorate the economical aspect of the power plant, while adhering to stringent emission standards. These control techniques explicitly incorporate the available process knowledge, which is represented in terms of an available mathematical model, used by the controller to compute the best control actions to fulfill the multiple conflicting goals in the plant. Therefore, mathematical modeling will be carried out to derive a suitable dynamic model of the power plant. The model is extended by designing a soft-sensor which estimates the energy content of fuel mix.