| 基于支持向量机的无参时变过程控制方法研究 |
| Research on adaptive support vector machine for time-varying process control |
| 摘要点击 293 全文点击 0 投稿时间：2020-07-28 修订日期：2021-01-22 |
| 查看/发表评论 下载PDF阅读器 |
| 中文关键词 时变过程; 过程控制; 支持向量机 |
| 英文关键词 time-varying process; process control; support vector machine |
| 基金项目 国家自然科学基金项目（面上项目，重点项目，重大项目） |
| 作者 | 单位 | 邮编 | | 何桢 | 天津大学 | 300072 | | 张静静 | 天津大学 | | | 何曙光* | 天津大学 | 300072 |
|
| 中文摘要 |
| 在实际生产中,生产过程通常表现为缓慢的时变模式,即过程的均值或方差发生缓慢的规律性或趋势性变化.而传统控制图是基于历史数据构建的,因此不能有效地对时变过程进行监控.本文提出了基于支持向量机的时变控制图,能够根据生产过程的变化及时更新分类器模型和控制界限,并根据第二阶段的数据构建动态分类边界,以实现对时变生产过程进行有效监控.同时通过添加时间权重因子和停止更新准则,实现对模型更新速度和更新区域的调节,使所提的时变控制图能够适应不同的时变模式.最后,通过模拟数据和青霉素发酵数据进行验证,证明所提的时变控制图能够对时变过程进行有效的实时监控. |
| 英文摘要 |
| A manufacturing process often shows a slow time-varying pattern, that is, the mean or variance of the process changes slowly in a trend or regularity. The traditional control chart is built with historical data, so it can’t effectively monitor the time-varying process. A time-varying control chart based on support vector machine is proposed to monitor the time-varying process, which can update the classifier model and control boundary in time according to the change of the process, and build a dynamic classification boundary according to the data of the second stage, so as to realize the effective monitoring of the time-varying process. At the same time, time parameters and update parameters is proposed to adjust the update speed update area of the model, so that the time-varying control chart can adapt to different time-varying modes. At last, through the simulation data and penicillin fermentation data, it is proved that the proposed time-varying control chart can effectively monitor the time-varying process in real-time. |
| 关闭 |
|
|
|
|
|
