International Journal of Information Technology and Applications, Vol. 1, No. 2, pp. 51-57, June 2024.
Abstract: The chromatographic separation process is a continuous and discrete mixed system. It consists of a fixed bed of chromatographic columns or similar structures, with components separated by valve switching. System parameters need to be controlled during the separation process, making it susceptible to external interference. In actual production, deviations from optimal operating conditions often occur, hindering the full utilization of the SMB chromatographic separation process’s capabilities. To unleash the potential of separation systems, results based on simulation models are crucial. This paper digitizes the simulation process, then utilizes Long Short-Term Memory (LSTM) neural networks to train simulation data, understanding the impact of observation parameters on outcomes. By identifying trends in complex parameter changes through LSTM learning, it establishes a foundation for verifying system controllability.