基于广义预测一碳饱充过程的控制研究

Study on Control of First Carbonation Via Generalized Predictive Control Algorithm

糖厂澄清工段一碳饱充过程是一个具有强非线性、多输入等特点的复杂系统,存在很多不确定因素,所以一直以来都没有很好地解决澄清过程的稳定控制问题。针对上述问题,采用数据的建模方法,对糖厂澄清工段一碳饱充进行BP神经网络建模,然后把模型作为广义预测控制的预测模型,将广义预测控制算法对糖厂澄清工段一碳饱充进行优化控制。在Matlab上进行仿真,结果表明,BP神经网络的广义预测控制算法具有鲁棒性强、控制精度高等优点,并使糖厂澄清工段一碳饱充系统输出很快稳定到设定值,实现了优化控制,减少能耗。

The first carbonation in the clarifying process of sugar cane juice is a complex system which has a strong nonlinear constraint. Because of these uncertain factors, there has not been a very good solution on the stability control of clarifying process. The idea is building a BP neural network model of the first carbonation based on data method. This model was consided as the forecast model of generalized predictive control, and a generalized predictive control algorithm was applied to the first carbonation for the optimal control. The result of Matlab simulation shows that generalized predictive control algorithm based on BP neural network can implement the optimal control of the first carbonation with strong robustness and high control precision.