混合高斯过程回归模型在铁水硅含量预报中的应用 = Composite gaussian process regression model and its application to prediction of silicon content in hot metal

• Main Authors: 任江洪 Ren, Jiang-Hong, 陈韬 Chen, Tao, 曹长修 Cao, Chang-Xiu
• Other Authors: School of Chemical and Biomedical Engineering
• Format: Article
• Language: Chinese; Chinese
• Published: 2015
• Subjects: DRNTU::Science::Medicine::Biomedical engineering
• Online Access: https://hdl.handle.net/10356/107138; http://hdl.handle.net/10220/25398
• Institution: Nanyang Technological University

为了提高基于高斯过程回归的软测量模型的预测精度,提出了一种混合高斯过程回归模型。该模型将高斯过程回归模型预测输出值的方差及其分布作为主要考虑因素,对多个高斯过程回归模型的输出值进行组合输出,获得了比单个高斯过程回归模型更高的预测精度和更强的模型鲁棒性。将该模型实用于高炉铁水硅含量预报模型的建模,获得了比使用单个高斯过程回归模型建模时更好的应用效果。In order to increase the predictive precision of gaussian process regression based soft sensor, a composite gaussian process regression model is proposed. This model combines the outputs of several gaussian process models as the output according to the variances and the distribution of the outputs, which results in higher prediction accuracy and higher robustness than the single gaussian process model. The proposed composite gaussian process regression model is successfully applied to the prediction of silicon content in hot metal.