摘要：SMB chromatography (simulated moving bed chromatography) is a separation science. T he soft measurement modeling approach of SMB chromatographic separation process based on extreme learning machine (ELM) with variable excitation function was introduced as a method for determining component fineness and harvest of extract and residue. The soft-sensor model's supplemental variables, as well as the crucial economic and cultural forecasting indices, were chosen after reviewing the technique of SMB chromatographic separation. Second, five excitation functions (Sig Sin, Har dlim,Tribas, and Radbas) are being used in the ELM neural network to generate the soft- sensor model, and the quantity of neurons in the hidden stratum of the ELM is determined.The simulation experiments consequences indicate that ELM neural network can availably achieve the accurate forecasting of key economic and technical indicators and fulfill the real-fime,efficient and robust operation of SMB chromatographic separation process.
关键词：SMB chromatographic separation;soft- sensing;extreme learning machine;excitation function
