基于梯度提升回归树算法的生活用纸皱纹等级软测量模型

Soft Measurement Model Based on Gradient Boosting Regression Tree Algorithm forCreases Level of Tissue Paper

关键词:  起皱  皱纹等级  软测量  梯度提升回归树算法

Key Words:creping  creases level  soft measurement  gradient boosting regression tree algorithm

摘要:皱纹等级是衡量生活用纸质量的重要指标之一。然而，工业生产过程中缺少皱纹等级的实时在线测量方法。为了解决上述问题，本研究通过实验对影响生活用纸皱纹质量的因素进行了分析。利用梯度提升回归树算法，对影响皱纹等级的表面粗糙度、皱纹深度、皱纹频率3个主要指标进行了建模，并通过预测这3个指标实现对皱纹等级的在线实时软测量。通过对比工业实测数据，发现该模型对表面粗糙度、皱纹深度、皱纹频率预测精度较高，测试数据的平均相对误差均小于5%。该模型解决了生活用纸皱纹等级在线软测量的问题，对生活用纸生产过程的质量控制提供了新的方法和依据。

Abstract:Creases level is one of the most important indicators to measure the quality of tissue paper. However, there is a lack of the real-time on-line measurement method of creases level in production. In order to address this issue, this paper analyzed the factors affecting the creases level of tissue paper by experiments. A soft measurement model of paper roughness，creases amplitude，creases frequency was established using the gradient boosting regression tree algorithm. And a real-time online soft measurement of creases levels could be realized by predicting these three indicators. In comparison with the industrial real data, it was found that the model had higher prediction accuracy for surface roughness, amplitude and frequency of the creases. The average relative error of the testing data was less than 5%. This model solved the problem of online soft measurement of creases level. It provided a new method and basis for the quality control of the tissue production process.