A nonlinear system modelling approach to industrial cane sugar crystallization

Guancheng Lu, Yanmei Meng, Jian Chen, Zhihong Tang, Xiaochun Wang, Xian Yu 

College of Mechanical Engineering, Guangxi University, No. 100, Daxue Road, Nanning, China

Cane sugar crystallization is a non-linear process where multiple control parameters are involved, which makes it rather difficult to reveal its internal mechanism by mechanism modelling. Derived from variants of standard support vector machine method, an online control system modelling method based on multi-input and multi-output proximal least square support vector machine is proposed to be applied in sugar crystallization process. This method takes multiple process control parameters as the input and output of machine learning algorithm, through which the inherent law between key and auxiliary parameters in the sugar crystallization process is established. The ultimate goal is to control the sugar crystallization process automatically. The experimental results show that the accuracy rate of the model output is 95%.