Li Xiaobing, Zhu Wei, Liu Jiongtian, Zhang Jian, Xu Hongxiang, Deng Xiaowei
a National Center for Coal Preparation and Purification Engineering Research , China University of Mining and Technology , Jiangsu 221116 Xuzhou , People's Republic of China.
b Sinopec Petroleum Engineering Corportation , Shandong 257026 , Dongying , People's Republic of China.
Environ Technol. 2016;37(7):785-94. doi: 10.1080/09593330.2015.1085098. Epub 2015 Oct 20.
The present work has been carried out to investigate the effect of process variables on gas holdup and develop an empirical equation and a neural network model for online process control of the gas holdup based on the operating variables. In this study, the effect of process variables (nozzle diameter, circulation pressure, aeration rate, and frother dosage) on gas holdup in a cyclone-static micro-bubble flotation column of an air/oily wastewater system was investigated. Gas holdup was estimated using a pressure difference method and an empirical equation was proposed to predict gas holdup. A general regression neural network (GRNN) model was also introduced to predict gas holdup for the cyclone-static micro-bubble flotation column. The predictions from the empirical equation and the GRNN are in good agreement with the experiment data for gas holdup, while the GRNN provides higher accuracy and stability compared with that of the empirical equation.
开展本工作是为了研究工艺变量对气体滞留率的影响，并基于操作变量开发一个经验方程和一个神经网络模型，用于气体滞留率的在线过程控制。在本研究中，研究了工艺变量（喷嘴直径、循环压力、曝气速率和起泡剂用量）对空气/含油废水系统的旋流-静态微泡浮选柱中气体滞留率的影响。采用压差法估算气体滞留率，并提出了一个经验方程来预测气体滞留率。还引入了一个广义回归神经网络（GRNN）模型来预测旋流-静态微泡浮选柱的气体滞留率。经验方程和GRNN的预测结果与气体滞留率的实验数据吻合良好，而GRNN与经验方程相比具有更高的准确性和稳定性。
