Modelagem e controle preditivo dos queimadores de uma caldeira

• Ano de defesa: 2019
• Autor(a) principal: Barbarioli, Guilherme Lima
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Engenharia Elétrica; Centro Tecnológico; UFES; Programa de Pós-Graduação em Engenharia Elétrica
• Palavras-chave em Português: Sistemas híbridos; Controle; MLD; HYSDEL; Kalman; Hybrid system; Control; subject.br-rjbn; Engenharia Elétrica
• Link de acesso: http://repositorio.ufes.br/handle/10/13531

Environmental sustainability has been a main topic leading industries to a better energy efficiency. Basic industries such as the metallurgic sector, applied many improvements in the last 50 years, mainly through new control and process techniques contributing to a outstanding 60% increase in energy efficiency, throughout all steel plants processes. However, those changes do not involve managing steel mill byproduct gases. The major issue that need to be addressed is the management and control systems that handle the destination of those gases. Mostly, those gases are used to generate electric energy and also to heat up some internal steps in the metal making process. To optimize the gas usage, the management strategies must include some mathematical optimizing models, to manage and predict the best route and time to consume those gases. Subjected to that management tool there must be a set of advanced controls that manage optimally the permissions and regulatory control to achieve the goals set by management algorithm. On this paper is proposed a hybrid control system based on furnace model of a boiler that is used to supply vapor for thermoelectric power plant. The main objective is to track the set-point path without any offset. Targeting this performance is proposed a hybridmodel using MLD (Mixed Logical Dynamics) formulation, utilizing HYSDEL(Hybrid System DEscription Language) language tools. This mathematical model attach many of the start-up and shut-down procedures, physical operation limits and also the switch affine systems that compose the furnace model. The control system use an on-line quadratic mixed-integer optimization with a receding control horizon, to predict the best control actions. To estimate some of the non-measured variables it was built a switched Kalman f ilter, allowing the control system to reject white noise and also constant disturb signal such as the valve grasp issue. In order to evaluate the control system performance, it is executed five case studies, including a comparison with a PI (Proportional Integral) controller.