Research On Computer-Aided Design And Application In Heavy Oil Catalyst

With the improvement of the industrialization, Catalyst plays great roles in the use of resources. But whether can design the high-effect catalyst formula, it has close something with methods. Unfortunately, by using traditional method, it will cost a lot of time to design a good formula. By surveying the development of the computer aided catalyst formula optimization and the related region and reviewing a great deal of catalysts developed by the former researchers, this paper formed a solution to the catalyst computer aided design and optimization, which includes experiment design method, mathematic modeling method and optimization method.Through theoretic analysis, the uniformity aimed experiment design method times are less than others. So, this paper formed a method to get the wanted uniform design by computer. Because the effect of component in catalyst is very complex and no explicit function expression can be gotten before build up the relationship between the recipes of catalyst and its performance. ANN can be a universal function approximation tool if an ANN has a structure of three layers: input layer, hidden layer and output layer. This paper chooses sigmoid function as activation function. Compared with traditional BP algorithm, momentum BP algorithm, Levenberg-Marquardt algorithm could get faster convergence rate. As to the optimal algorithm, the genetic algorithm is adopted. It can carry on the operation on a population composed by an encoding collection of variables and realize an effective search and the optimal solution of overall population. Based on the testing results, a neural network, which contained 4, 6 and 1 neurons in input layer, hidden layer and output layer respectively, was determined as the best network to establish the catalytic reaction model.Finally, some problems on the catalyst formula method and the steady-state optimization are put forward that are remained to explore and study.