基于梯度提升决策树的湿法磷酸过程软测量建模

Soft sensor modeling of wet process phosphoric acid process based on gradient lifting decision tree

在磷酸生产过程中,某些关键的质量指标无法直接测量,可能对生产过程和产品品质造成不利影响。介绍一种新的集成学习算法,并开发了一种基于梯度提升决策树的机器学习软测量技术方法,以处理磷酸反应槽生产过程中的强非线性和动态性。通过湿法磷酸反应槽装置基准验证了改进的建模方法的有效性。经磷化工生产实际应用表明,该方法适用于磷化工生产过程建模。

In the production process of phosphoric acid,some key quality indicators cannot be directly measured,which may have adverse effects on the production process and product quality.The article introduces an ensemble learning algorithm and develops a machine learning soft sensing technology method based on gradient lifting decision tree to deal with strong nonlinearity and dynamics in the production process of phosphoric acid reaction tanks.The effectiveness of the improved modeling method was verified through the benchmark of the wet process phosphoric acid reaction tank device.Then,it is applied to practical cases of phosphorus chemical production,indicating that this method is particularly suitable for modeling phosphorus chemical production processes.