Modelling study of p-xylene oxidative side-reaction based on Bayesian filtering
Liu Ranran1, 2, Li Zhengming1
COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(2) 95-98
1 School of Electrical and Information Engineering, Jiangsu University, Zhenjiang 212013, China
2 Changzhou Liu Guojun Vocational Technology College, Changzhou 213004, China
The combustion loss model of HAC and PX was established based on Bayesian filtering. Moreover, this neural network model regarded the main adjustable process parameters (residence time, the concentration of Co, Mn, Br, reaction temperature and solvent ratio) as the independent variables and the total content of CO and CO2 as the dependent variable. The simulation results show that the network is well performed. The effects of process operating parameters on HAC and PX combustion loss are analysed based on the model. Wherein, the effects of residence time, reaction temperature on the combustion loss are bigger, while the effects of other factors are relatively small.