A New Method For Modeling And Optimization Of Gasoline Yield Of MIP Process

The maximizing iso-paraffins(MIP) catalytic cracking process belongs to family process of catalytic cracking. The reactor in MIP has two continual parts, which have different diameter. In this way, the quality of oil products is improved. This kind of process can maximize iso-paraffins and reduce alkene in gasoline while not changing the gasoline octane number.In this paper, the gasoline yield prediction model is established and the improved genetic algorithm is used to optimize gasoline yield. The prediction model is based on generalized regression neural network (GRNN) and adaboost algorithm. There are 16 input variables and one output variable in this model. The input variables reflect properties of raw oil, catalyst and operation condition. The output variable is gasoline yield. The result shows that gasoline yield predicted by model has mean square error (MSE) of 0.44. The improved genetic algorithm is used to find out highest gasoline yield and corresponding values of four operation variables when the properties of raw oil and processing capacity don’t change. The four operation variables are:ratio of cycle, ratio of catalyst to oil, outlet temperature of the second stage of reactor and preheating temperature of raw oil. The precision of prediction model is good. The optimization results fit in production experience.The prediction result of gasoline model and optimization results by improved genetic algorithm confirm the effectiveness of the application of neural network and intelligent optimization algorithm in control and optimization of catalytic cracking. This work provides beneficial experience for further industrial application.