Research on methanol distillon agent model based on Aspen Plus

• 1、Chemistry and Chemical Engineering School, Chongqing University, Chongqing 401331

Abstract：There are many variables in the chemical production process that are difficult to measure directly, and in general the variables that are difficult to measure can be simulated using simulation software to produce the corresponding simulation results for the operator to regulate the production. However, common process simulation software has problems such as slow calculation speed and difficulty in convergence. In recent years, neural networks have developed rapidly. Neural networks have the advantage of being able to fully restore complex non-linear relationships and converge quickly, so they can effectively solve the problems that exist in conventional process simulation software. This paper proposes the use of actual chemical production DCS data as input, Aspen Plus simulation data as output, and the use of feedforward neural networks to establish the agent model, to solve the problem of the slow running speed of the traditional process simulation software calculatio。

Keywords： Aspen Plus Neural Networks Data-driven Agent Model Methanol Distillation