Exploring the Effect of Reaction Conditions on the Yield of C4 Olefins Based on Neural Networks and Improved Simulated Annealing

In this paper, a mathematical model was developed to investigate the catalyst combination and reaction temperature in the chemical reaction by rational quantification of experimental parameters with the objective of maximum C4 olefin yield in the context of C4 olefin preparation by ethanol coupling. Based on the existing data, a BP neural network model was established to obtain the BP neural network relationship function with Co loading, the mass of HAP, the mass of Co/SiO2 and ethanol concentration as input samples, and ethanol conversion and C4 olefin selectivity size as output samples. The simulated annealing algorithm was improved by setting the generation rules of the new solution and defining the penalty function. With the maximum C4 olefin yield as the solution objective, the catalyst combination and temperature of 97.4mg 2.5wt% Co/SiO2- 81.4mg HAP-ethanol concentration of 1.44ml/min and 499°C were obtained by iterating the improved simulated annealing algorithm. The relationship between various parameters was obtained by finding a suitable model for mathematical analysis. The theory is used to guide the chemical practice, to reduce the waste of various resources in the experimental process, and to promote the development of the experiment.