The main objective of this study was to optimize the design of a production process for the preparation of C4 olefins from ethanol. Firstly, the data were preprocessed to investigate the association between temperature, ethanol conversion, and C4 olefin selectivity for various catalyst combinations using polynomial fitting methods based on data distribution patterns. Secondly, SVM regression, Gaussian process regression, and BP neural network regression models were used to investigate and select the best models for ethanol conversion and C4 olefin yield for different catalyst combinations and temperatures. Finally, neural network and particle swarm optimization algorithms were used to derive the optimal catalyst combination and temperature to maximize C4 olefin yield. The use of neural networks and particle swarm optimization algorithms proved to be effective in optimizing the reaction conditions for the production of C4 olefins from ethanol.

本研究的主要目的是优化乙醇制备 C4 烯烃的生产工艺设计。首先，使用基于数据分布模式的多项式拟合方法对数据进行预处理，以研究各种催化剂组合的温度、乙醇转化率和 C4 烯烃选择性之间的关联。其次，利用SVM回归、高斯过程回归和BP神经网络回归模型研究并选择不同催化剂组合和温度下乙醇转化率和C4烯烃收率的最佳模型。最后，使用神经网络和粒子群优化算法推导出最佳催化剂组合和温度，以最大化 C4 烯烃产量。