两段法固定床甲醇制芳烃产物预测多元非线性回归模型

针对甲醇制芳烃机理建模假设多、产物预测过程复杂、计算成本高等问题，提出了一种多元非线性回归分析的数据驱动建模方法。在两段法固定床反应系统中研究了反应压力、甲醇体积空速、一段中心温度、二段中心温度、装置运行时间、累计甲醇进料量及其交互作用对芳烃产物的影响。利用最小二乘法进行参数估计，建立四个六元二次非线性产物分布回归模型。测试结果表明，测试样本的总体决定性指标{\rho }^{\mathrm{2}}为0.9576，均方误差MSE为0.0037，相比于传统的动力学模型，具有产物预测精确度高、计算量小以及泛化性强的特点；空速和反应压力的交互作用对产物中芳烃选择性影响显著，在空速增大的情况下，产物中最大芳烃占比对应的压力峰值也随之增高。

Multivariate nonlinear regression model of methanol to aromatics by two-state fixed bed for product prediction

Aiming at the problems of multiple assumptions about the mechanism modeling of methanol-to-aromatics, complex product prediction process, and high computational cost, a data-driven modeling method based on multiple nonlinear regression analysis was proposed. The influence of reaction pressure, methanol volume space velocity, central temperature, running time, effects of cumulative methanol feed and its interaction on aromatic products is studied in the two-stage fixed bed reaction system. The least square method is used to estimate the parameters and four regression models of quadratic nonlinear product distribution are established. The test results show that the overall decisive indicator of the test sample is 0.9576 and the mean square error MSE is 0.0037, which shows the high prediction accuracy of the product, time saving and strong generalization compared with the traditional kinetic model; The interaction between air speed and pressure influences the hydrocarbon selectivity significantly, and the pressure peak corresponding to the maximum hydrocarbon proportion in the product also increases with the airspeed increases.