关键词: 非线性量化, 小脑模型神经网络, 溶出预脱硅, 循环母液, 时间序列预测

Abstract: The Cerebella Model Articulation Controller（CMAC） based on nonlinear quantization to adjust the concept mapping is presented.The CMAC based on nonlinear quantization improves the speed and accuracy of calculation to meet the complex and dynamic demand in the nonlinear environment for real-time control.Considering the demand of the process optimization of digesting pre-silicon systems，a proportion mixture time series prediction model of digesting recycled liquor by CMAC based on nonlinear quantization is presented to forecast the quantity of recycled liquor accurately and fast.The quantity of recycled liquor is optimized in this application.Industry test shows that the accuracy and rapid of the time series prediction model has obvious advantages than other models.And the model has been applied to certain alumina plant to optimize dynamically the recycled liquor quantity to save the cost of production and achieve obvious economic benefits.

Key words: nonlinear quantization, Cerebella Model Articulation Controller（CMAC）, digesting pre-silicon, recycled liquor, time series prediction