Abstract：For the sake of the concentration of production cost and maintenance cost, polynomial soft sensor model of the overflow particle size of cyclone based on AdaBoost is applied to solve it. Minimum mean square error is objective function to establish a polynomial model and strengthens the training sample of the poor performance to improve the generalization and the precision of the model. In the end many models are integrate as the final model according to the different weights. The simulation results of field data and the model online application effect show that the model can more accurately predict the cyclone overflow particle size, and ensure the qualification rate of the cyclone overflow particle size, and improve the separation efficiency, so the final model has a certain practicality.
李晓竹,解华. 基于AdaBoost的多项式旋流器溢流粒度软测量模型[J]. 计量学报, 2016, 37(3): 275-278.
LI Xiao-zhu,XIE-hua. Polynomial soft sensor model of the overflow particle size of cyclone based on AdaBoost. Acta Metrologica Sinica, 2016, 37(3): 275-278.
［1］庞学诗.水力旋流器工艺计算［M］.北京:中国石化出版社,1997.
［2］Joseph B,Brosilow C B. Inferential Control of Processes:Part I.Steady State Analysis and Design［J］.AIChE Journal,1978,24(3):485-492
［3］孙万田, 邬齐斌.密炼机的推断控制系统［J］.化工自动化及仪表,2001,28(5): 44-46.
［4］李海青,黄志尧.软测量技术原理及应用［M］.北京:化学工业出版社,2000.
［5］任若恩,王惠文.多元统计数据分析［M］.北京:国防工业出版社,1997.
［6］田慧欣,王安娜.基于增量学习思想改进的AdaBoost建模方法［J］.控制与决策,2011.
［7］蒋艳凰,赵强力.机器学习方法［M］.北京:电子工业出版社,2009.
［8］Vapnik V.统计学习理论的本质［M］.张学工,译. 北京:清华大学出版社,2000.
［9］张晓东,王伟,王小刚.选矿过程神经网络粒度软测量方法的研究［J］.控制理论与应用,2002,19(1): 85-88.
［10］丁进良,岳恒,齐玉涛,等.基于遗传算法的磨矿粒度神经网络软测量［J］.仪器仪表学报,2006,27(9): 981-984.
［11］龚纯,王正林.精通MATLAB最优化计算［M］.北京:电子工业出版社,2009:270-312.
2016, Vol. 37 
