On-line/off-line optimization of complex processes using a linguistic self-organized optimizing control scheme

Abstract

Abstract Chemical processes usually have a complicated structure and are subjected to unknown long term and short term disturbances. Many process optimization schemes were developed under the assumption of the existence of a model that can predict the process steady state and/or dynamics. However, such a model may not be precise nor even exist. A rule based methodology without the need of such a model is hence developed in this work. The optimizing control scheme developed here is actually a combination of a modified version of the linguistic control scheme developed by Procyk and Mamdani and the concept of reduced gradient optimization. It is assumed that the objective function and constraints of the optimization problem are functions of manipulated variables and measurements only. The optimizing control algorithm is also formulated into a look-up table. The rule-based optimizing control scheme possesses three important properties: adapting, learning and decision-making. Some numerical examples of multivariable nature and constrained situations are solved by this approach and compared with well-known approaches. It is shown that the method is superior to the previous methods. A process system consisting of a series of CSTR has been simulated as an illustrative example. The simulation results show that the optimizing control scheme is promising and should be of interest to process industry.