|
|
基于梯度增强决策树算法的纸张质量软测量模型 |
Gradient Boosting Tree Algorithm Based Soft Measurement Model for Paper Quality |
收稿日期:2019-12-25 修订日期:2020-01-13 |
DOI: |
关键词: 数据模型 纸张质量 软测量 GBT |
Key Words:Data model Paper quality Soft sensing technology Gradient Boosting Tree algorithm |
基金项目:制浆造纸工程国家重点实验室开放基金(201830) |
|
摘要点击次数: 655 |
全文下载次数: 0 |
摘要:对纸张的关键物理指标如抗张强度、柔软度和松厚度进行了在线软测量,从而克服质检仪器下抽检以及结果反馈存在时间滞后的问题,为连续生产的优化控制提供实时决策。本研究提出了一种基于梯度增强决策树(GBT)的纸张质量软测量模型,该方法可在线软测量生产纸张的关键物理指标如抗张强度、柔软度和松厚度。结果表明,采用GBT进行纸张质量软测量时,得到抗张强度的平均相对误差为7.21%,柔软度的平均相对误差为7.38%,松厚度的平均相对误差为3.5%,满足质检误差要求,可为稳定产品质量、生产过程优化及降低生产成本提供参与。 |
Abstract:Online soft measurement of key physical indicators such as tensile strength, softness and bulk is carried out to overcome the problem of time lag in sampling inspection and feedback of quality inspection instruments, and to provide real-time decision-making for continuous control of continuous production. This paper proposes a paper quality soft measurement model based on Gradient Boosting Tree (GBT), which can softly measure the key physical properties of paper such as tensile strength, softness and bulk. The results show that when GBT is used for soft measurement of paper quality, the average relative error of tensile strength is 7.21%, the average relative error of softness is 7.38%, and the average relative error of bulk is 3.5%, which meets the requirements of quality inspection error. Participation can be provided to stabilize product quality, optimize production processes and reduce production costs. |
HTML 查看/发表评论 下载PDF阅读器 |
|
|
|