Research On Neural Generalized Predictive Control For Optimal Control In The Clarifying Process Of Sugar Cane Juice

Clarifying process of sugar cane juice is a very important craft of the sugar factory. Clarifying process of sugar cane juice is a complex system which has a strongly nonlinear constraint, multiple input characteristics. Because of these factors, the stability control problem of clarifying process has been not very good solution. Looking for a new control method to realize the optimal control in clarifying process of sugar cane juice is a significant issue.Generalized predictive control is widely used in the industrial process control, and made the obvious economic benefits because Generalized predictive control need less certain parameters, and has a few the amount of calculation, the design of the control system is flexible and convenient.Firstly, the whole process of clarifying process of sugar cane juice and all kinds of factors are analyzed. Considering the complexity of the whole system, The whole system is divided into several small systems, and are modeling. The first carbonation and the second carbonation in the clarifying process of sugar cane juice are selected, and the BP neural network is applied to the model of the two processes. Secondly, the model is as the controlled object of GPC, and applies the BP neural network to the GPC s predictive model of the controlled object which is made up of the online identification and predictor based on BP neural network. Taylor formula is used to linearize the nonlinear model, and get a linear model. Finally, generalized predictive control algorithm is applied to the linear model. The control process of the first carbonation and the second carbonation is divided into two parts in the paper. The first part is that when the flow is constant, the output value is controlled from the initial value to the set value and finally the output value becomes stable. Another part is that when the flow is not constant and the first part has been stable, the changed output value is controlled to the set value. The simulation results show that GPC combined with neural network is very effective to the control of the two processes in the clarifying process of sugar cane juice, and get the control target controlled in the set value range and the controlled quantity is changed in the required range, no matter the flow is constant, or changed.