Design of regulative control laws for stabilization of regenerator temperature in adiabatic Fluid Catalytic Cracking Units is performed. This is achieved despite the fact that kinetic terms and heat generation are, commonly, unknown. This article presents a control law that makes use of on-line discrete measurements of temperature to estimate the heat of reaction, obtained from a neuralestimation algorithm. The controller design is similar to discrete PID controllers. Some new tuning rules are given in terms of closed loop and measured characteristic times. In order to obtain the simplest neural network and the minimum number of parameters to be adjusted, only one perceptron is used. Moreover, the resulting control law has physical meaning and can be interpreted as a standard PID controller with time varying gain. Performance of the neural-estimation technique is analyzed under closed-loop behavior using numerical simulations considering disturbances to the process and set point changes. The performance obtained when using this new control is compared with the one obtained by using a PID controller tuned following the IMC rules.
- Model uncertainties,
- Temperature regulation
Available at: http://works.bepress.com/ricardo_aguilar_lopez/15/