论文部分内容阅读
现代工业生产的复杂化使得候选辅助变量增多,大量的输入变量会使软测量模型过拟合,影响模型的预测效果。针对这个问题,提出了一种蒙特卡洛无信息变量消除结合遗传算法偏最小二乘(MC-UVE-GA-PLS)选择辅助变量的方法。该方法在运用GA算法搜索最优变量子集之前,采用MC-UVE方法消除与模型不相关的变量,使GA算法能有效地搜索出对响应变量预测贡献最大的变量子集。用本文提出的方法建立了工业精馏塔浓度软测量模型,仿真结果表明本文提出的辅助变量选择方法不仅能提高模型的预测能力,而且能简化模型的复杂性。
The complexity of modern industrial production leads to the increase of candidate auxiliary variables. A large number of input variables will over-fit the soft-sensing model and affect the prediction effect of the model. To solve this problem, this paper proposes a Monte Carlo non-information variable elimination method combined with Genetic Algorithm Partial Least Squares (MC-UVE-GA-PLS) to select auxiliary variables. Before GA algorithm is used to search the optimal variable subset, this method uses MC-UVE method to eliminate the variables that are not related to the model, so that the GA algorithm can effectively search for the subset of variables that contribute most to the prediction of response variables. The method proposed in this paper is used to establish the concentration soft sensor model of industrial distillation column. The simulation results show that the proposed auxiliary variable selection method can not only improve the predictive ability of the model, but also simplify the complexity of the model.