The Application Of Advanced Control In The Pilot Production Of Gas-Phase Polyethylene

In recent years, Advanced Control has many applications in the petroleum and chemical industries. The implementation of Advanced Control and online optimization technology makes a more stable process control and brings significant economic benefits. So, the Advanced Control applied in an actual gas-phase pilot production is the main content of this paper.This paper mainly studied on the following aspects:This paper analyzed the original temperature control system of the gas-phase polyethylene reactor to find the reasons to make the large temperature fluctuation exist in the temperature control system. There were two main reasons including:Much disturbing existed in the temperature control system, polymerization reaction is nonlinear and it has the characterization of large hunting. For these reasons, the improved strategy was determined:The triple temperature cascade control system replaced the existing dual temperature cascade control system. The improved strategy was performed in the distributed control system, and the better control performance was gotten.This paper also researched the soft-sensing for polyethylene quality indicators including Melt Index and Density, and monitored the polyethylene production. The process data used in the soft-sensing and system modeling was preprocessed by removing abnormal data and smoothing data before modeling. By analyzing the system characteristics, the neural network was determined to make a model. According to reaction mechanism, it determined the variables of the inputs, outputs and intermediate state quantities in the neural network model. The model's type and structure were chosen reasonably, at the same time the parameters in the model were set. It selected Legendre polynomial as the excitation function of the neural network and least-square method as the training algorithm in the model. Three types of network structure were built for the system model which had been trained, tested and comprised.Optimal control theory and gradient algorithm were used to research the simulation of advanced control on polyethylene melt index and density. According to reaction mechanism, the control variables in the control of polyethylene quality indicators were determined. On the basis of the model built for the soft-sensing for polyethylene quality indicators, the optimal control algorithm was deduced. The actual system was simulated by adding white noise to the outputs of the neuronal network of the soft-sensing on the process of simulation.