Dynamic simulation of distillation column based on Levenberg-Marquardt algorithm

Wang Xingjun,Hu Lishun,Yu Guangsuo,Wang Fuchen,Yu Zunhong
DOI: https://doi.org/10.3969/j.issn.1001-4160.2006.12.031
2006-01-01
Abstract:Dynamic simulation of distillation column based on the core of nerve network theory is studied. The optimizational nerve network Levenberg-Marquardt algorithm is provided for dynamic simulation of methanol distillation column on the base of Newton global optimizational method and multilayer feedforward BP nerve network. The results show that the feed components affect contents of gas components in the top of column and contents of liquid components in the bottom of column. Cas flow rate in the top of column and liquid flow rate in the bottom of column both change when the system of steady state is broken for the changing feed component. And, new e-quilibrium system comes into being in the end. The little changes of feed components and ratio of circumfluence largely affect on the e-quilibrium state of methanol distillation column.
What problem does this paper attempt to address?