Smith predictor-based controllers for high-order dead-time processes

• Ano de defesa: 2023
• Autor(a) principal: Pereira, René Descartes Olimpio
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Palavras-chave em Português: Preditor de Smith; . Processos com atraso de transporte; Modelos de ordem elevada; Zeros de fase não-miníma; Controle feedforward; Perturbações mensuráveis; Incubadora neonatal; Smith predictor; Dead-time processes; High-order models; Non-minimum phase zeros; Feedforward control; Measurable disturbances; Newborn intensive care unit
• Link de acesso: http://repositorio.ufc.br/handle/riufc/76049

This work studies and proposes controllers based on the Smith predictor (SP) for both highorder dead-time processes and processes with measurable disturbances. Relevant SPbased controllers from the literature are studied, and their main characteristics are highlighted. A control structure suitable for high-order dead-time processes with nonminimum phase (NMP) zeros is proposed, namely, the simplified filtered SP (SFSP). The proposed strategy uses a state-space formulation and can be applied in a unified manner to stable, unstable, and integrating linear dead-time processes of any order. In general, dead- time compensators (DTCs) first predict the process output with zero error at a steady state, and then a primary controller with an integrator is designed based on the delay-free model. The main advantage of the proposed structure is that the primary controller is only a statefeedback gain with no explicit integrators. This leads to fewer parameters to tune and lowerorder filters, while a robustness filter is used to reject disturbances and guarantee zero error at a steady state. Simulation results show better or equivalent performance than other recently published works, even by keeping controller design simple. The basic fundamentals of feedforward control for measurable disturbances are presented, and important SP-based controllers for dead-time processes with measurable disturbances are studied. A feedforward extension for the SFSP for high-order dead-time processes to deal with measurable disturbances is also proposed. Compared to other control strategies, the main advantage of using the predictor approach is the ability to deal with processes with large dead time. The structure improves the disturbance rejection performance by feedforwarding the disturbance while maintaining the good robustness and noise attenuation properties of the SFSP. Simulation results also show better performance indices compared with other strategies from the recent literature. In addition, to show their effectiveness in a real process, the proposed strategies are applied to control the temperature in a newborn intensive care unit (NICU).