Nonlinear Model Predictive Control Of The Cane Sugar Crystallization Process

Since 1970's, model predictive control (MPC), which has advantage of good robustness and easily modeling, has widely applied in the industry. A great progress for theoretical study of linear MPC has reached while a uniform nonlinear MPC algorithm still hard to get for the complexity of nonlinear system.The process of sugar boiling is the process of sugar's crystallization. In this complicated chemical and physical process, both heat transfer and mass transfer occur. Since this process itself has many factors suffering disturbance and has great nonlinearity, so it is hard to control it with ordinary methods. In the paper, a process model is established based on mechanism and analytical relations of mass, energy and momentum and other physical properties in the sugar boiling process. A simulation has been taken to the oversaturation control of boiling sugar process by means of nonlinear MPC algorithms.After introducing basic theory of MPC and analyzing application of nonlinear MPC in cane sugar crystallization, then a MPC algorithm which based on B-P neural network model and genetic simulated annealing optimization is discussed.The main contents of this paper are as follows:1. A survey of the development and status of MPC is given. An introduction of related control strategies is also given;2. Systems analysis of the boiling crystallization process is made;3. A MPC algorithm is discussed;4. Both PID and based-on B-P neural network model and genetic simulated annealing optimization predictive control algorithms are simulated in Matlab and their results are analyzed.