Aspen Plus Pre-Verification Integrated with ANN Model for Online Composition Detection and Instantaneous Regulation of Olefin Gas-Phase Copolymerization System

Online monitoring of gas-phase composition is critical for regulating product performance in olefin copolymerization, but existing detection systems based on series-connected flowmeters face high costs and long delivery times of imported instruments, leading to significant risks of resource waste without prior feasibility verification. To address this gap, this study proposes a novel technical route integrating Aspen Plus pre-verification and an Artificial Neural Network (ANN) model. First, Aspen Plus was used to simulate the thermophysical properties (molar heat capacity and viscosity) of ethylene/propylene/hydrogen mixtures across ۴۸۸ operating conditions, thereby verifying the linear independence of two key flow-rate ratios derived from a thermal mass flowmeter, a Coriolis mass flowmeter, and a differential pressure flowmeter. This pre-verification assesses the feasibility of composition inversion before instrument procurement, avoiding unnecessary costs. Subsequently, an ANN model was constructed with temperature, pressure, and the two flow-rate ratios as inputs and gas molar fractions as outputs. The model achieved high precision on both simulated and experimental data: for the validation set of actual industrial data, R۲ values for ethylene, propylene, and hydrogen were ۰.۹۷۵, ۰.۹۷۸, and ۰.۹۶۷, respectively, with RMSE values less than ۰.۰۴۱. Finally, the ANN model was embedded into a DCS closed-loop control system, realizing program-controlled regulation of instantaneous copolymer composition with maximum relative deviations of ۴.۰۸% for key monomers. This study differs from previous works by emphasizing pre-verification of detection feasibility, reducing instrument procurement risks, and achieving precise control of instantaneous composition, providing a cost-effective and reliable solution for industrial online monitoring and process optimization.

کلیدواژه ها:

Aspen Plus pre-verification ، artificial neural network (ANN) ، online composition detection ، olefin gas-phase copolymerization ، feasibility verification ، Instantaneous composition control