基于域适应支持向量回归的磨机负荷软测量

Soft Sensor of Ball Mill Load Parameters Based on Domain Adaptation Support Vector Regression

在工况改变时,湿式球磨机的实时数据和建模数据分布不一致,不满足机器学习的数据同分布假设,采用传统软测量方法将导致软测量模型失准和性能恶化等问题。为此,引入域适应思想,提出一种基于域适应支持向量回归的软测量模型,实现多工况下湿式球磨机负荷参数的准确测量。首先对多工况数据进行预处理并提取频谱特征,然后利用目标域中少量带标签样本数据所蕴含的特征信息和知识结构,提升源域数据构建模型对目标域数据的适应程度,最后对磨机负荷参数进行回归预测。实验结果表明,该软测量方法显著优于现有方法,适用于多工况下的软测量建模。

When the working condition of wet ball mill is changed,the distributions of real time data and modeling data are inconsistent,and the independent and identically distributed assumption of machine learning would not be satisfied.In this case,the soft sensor model,obtained by the traditional soft sensor method,may be inaccuracy,which leads to the deterioration of the model performance.Therefore,by introducing the domain adaptation theory,this paper proposes a soft sensor model based on domain adaptive support vector regression strategy to achieve the precise measurement of wet ball mill load parameters under multiple working conditions.Firstly,data in multiple working conditions is preprocessed and spectrum features are extracted.For then,the feature information and knowledge structure of few labeled samples in the target domain data are used to improve the adaptation for target data of model constructed by source domain data.Finally,the mill load parameters are predicted by the regression model.The experimental results show that the proposed soft sensor method is significantly better than the existing method,and the proposed method is suitable for soft sensor modeling under the situation of multi working condition.