Research On The Products Quality And Yield Of Catalytic Reformer Prediction Model

The catalytic reforming is one take the gasoline (is mainly straight run gasoline) as the material production high octane value gasoline and the light aromatic hydrocarbon (benzene, toluene, xylene, are called BTX) the important refinery process, simultaneously also produces the considerable amount the by-product hydrogen. Catalytic reforming high petrol with an octane number of unleaded petrol is an important component. In developed countries the components in the car with gasoline, catalytic reforming gasoline accounts for about 25 percent to 30 percent. Benzene, toluene and xylene is a basic chemical raw materials, all over the world for more than half of BTX from the catalytic reforming, catalytic reforming is the cheapest source of hydrogen.The monitoring and control system is the integrated technology of the field bus technology, the computer technology, the communication, the data bank technology, the multimedia technologies, the control technology and the networking and so on. This paper has first analyzed the catalytic reforming the monitoring and control system's system request. Then separately to the monitoring and control system in the host computer and the slave computer design, specially to the host computer monitoring partial craft contact surface and the function contact surface's design, realizes, as well as the system redundancy, aspects and so on system maintenance have carried on the detailed introduction. Especially, this paper further researched the design and implementation of craft interfaces and function interfaces in the host computer and describes the realization of system redundancy, system maintaining and system network.Support Vector Regression is an important area of Statistical Learning Theory. It's basical theory is VC dimension theory and structure risk minition rule,and it prossesses goodcharacteristic and best generalization ability. RBF kernel function is selected by studying the primary of four kinds of SVM common used kernels. The parameter selection of SVM is studied, and some conclusions are gotten through experiments.Catalytic reforming product yield as well as product quality's predict that to entire build up enterprise's profit forecast and the production decision efficiency analysis have the very vital significance. This paper used support vector machines establishment catalytic reforming product yield and the product quality predictive models.But, how to select the model parameters SVM much suffered from man-made influence, and the choice process is more complex, moreover is very sometimes difficult to obtain the satisfactory solution. Considering the problem, on a basis of fully discussing SVM and PSO basic theory, the paper used a new technique based on PSO to select the model parameters for SVM. The simulation result indicated that this method surpasses the conventional routes in the prediction accuracy and the simulation time.