Análise de malhas de controle multivariável utilizando cálculo fracionário
Resumo
Resumo: Este estudo utiliza o cálculo fracionário para o desenvolvimento de controle PI para
sistemas multivariáveis de ordem inteira e fracionária. Em termos de processos
industriais, o controle proporcional integral (PI) é mais utilizado. Estudos apontaram
que o cálculo fracionário tem sido aplicado com sucesso tanto para a modelagem do
processo quanto para o projeto do controlador. Foi estudado o controle multivariável,
utilizando como base o modelo de ordem inteira de Wood e Berry (1973), de uma
coluna de destilação de água e metanol, e o modelo de ordem fracionária
apresentado por Li e Chen (2014), baseado em um experimento de controle de
temperatura em uma placa termoelétrica. Algoritmos genéticos foram implementados
para a sintonia do controlador com base na minimização de ISE e ITAE de todos os
laços, considerando a mudança de set-point em cada variável controlada. Como
critérios também foram considerados diminuição do overshoot, tempo de
estabilização e oscilação da malha. As simulações consideraram controladores
descentralizados, erros de medição aleatórios nas variáveis controladas e a
presença de desacopladores para diminuir a interação entre as malhas. O
controlador PI fracionário melhorou significativamente o desempenho dos sistemas
de controle em ambiente multivariável quando comparado ao controlador PI de
ordem inteira, reduzindo erros e tempos de resposta, além de permitir que a sintonia
seja ajustada mais facilmente de acordo com os critérios de desempenho definidos.
O melhor ajuste foi obtido através da função objetivo baseada no ITAE. Para o
sistema de Wood e Berry (1973) a função objetivo foi reduzida em 53,6% para uma
mudança de set-point da composição de fundo, em comparação com o controle
inteiro. Também foi alcançada uma melhoria de 58,9% ao comparar sistemas
desacoplados. Para o sistema de Li e Chen (2014) a função objetivo baseada no
ITAE foi reduzida em 63%, em comparação com o controle inteiro. Foram obtidos
61% de melhoria para o controlador fracionário ao comparar os sistemas
desacoplados. Os controladores fracionários foram estáveis com erros de até 10%
do valor da variável medida.
Palavras-chave: Controle de Processos, Controle PI Fracionário, Sistemas
Multivariáveis, Algoritmos Genéticos Abstract: This study uses the fractional calculus for the development of PI control for
multivariable systems of integer and fractional order. In terms of industrial processes,
the integral proportional control (PI) is more used. Studies have pointed out that the
fractional calculus has been applied successfully for both process modeling and
controller design. Multivariable control was studied using a Wood and Berry (1973)
model of a water and methanol distillation column as case basis, and the fractional
order model presented by Li and Chen (2014) based on a temperature control
experiment on a thermoelectric plate. Genetic algorithms were implemented to tune
the controller based on the ISE and ITAE minimization of all loops, considering the
set-point change in each controlled variable. As criteria were also considered
decrease of overshoot, stabilization time and oscillation. The simulations considered
decentralized controllers, random measurement errors in the controlled variables and
the presence of decouplers to decrease the interaction between the loops. The
fractional order PI controller significantly improved the performance of the control
systems in a multivariable environment when compared to the integer order PI
controller, reducing errors and response times, and allowing tuning to be adjusted
more easily according to the defined performance criteria. The best fit was obtained
through the objective function based on ITAE. For the Wood and Berry (1973)
system the objective function was reduced by 53.6% for a set-point change of the
bottom composition as compared to the integer control. An improvement of 58.9%
was also achieved when comparing decoupled systems. For the Li and Chen (2014)
system the objective function based on the ITAE was reduced by 63% compared to
the integer control. A 61% of improvement was obtained for the fractional controller
when comparing the decoupled systems. Fractional controllers were stable with
errors of up to 10% of the value of the measured variable. Keywords: Process Control, Fractional PI Control, Multivariable Systems, Genetic
Algorithms
Collections
- Dissertações [107]