论文部分内容阅读
提出一种将多变量非线性问题线性化的VDE-PSO-MLR的建模方法。该方法基于变量扩维、微粒群优化等手段选择扩维变量,在此基础上再建立拟线性的多元回归方程;并通过对所建立的各回归方程及其回归系数的显著性检验结果确定最佳回归模型。将该方法用于某炼油厂的汽油调合设计公式的挖掘,研究表明,与直接用自变量建立的线性回归方程以及二次回归方程相比,只有该方法建立的最佳模型方程和方程变量同时通过显著性检验。最后将最佳模型用于生产数据预测,计算调合汽油辛烷值测定值与预测值误差绝对值AE最大为0.185,符合AE≤0.3的要求。
A modeling method of VDE-PSO-MLR is proposed to linearize the multivariable nonlinear problem. The method selects the expansion variables based on variable expansion and particle swarm optimization, and then establishes the quasi-linear multiple regression equation based on this method. The regression equation and the significance test results of regression coefficients are established to determine the most significant Good regression model. The method is applied to the design of gasoline blending design in a refinery. The results show that compared with the linear regression equation and the quadratic regression equation established by independent variables, only the best model equation and equation variables At the same time through the significance test. Finally, the best model is used to predict the production data. The absolute value of the absolute value of the error between octane number and predicted value of blended gasoline is 0.185, which meets the requirements of AE≤0.3.