论文部分内容阅读
针对间歇过程数据存在动态变化特征,传统的支持向量数据描述(support vector data description,SVDD)方法很难实现实时在线状态监测的问题,提出一种基于滑动窗口的SVDD在线实时故障监测方法.通过采用适当大小的滑动窗口逐步更新当前子数据空间,建立SVDD子模型,从而实现在线实时故障监测.该方法不仅克服了过程数据非高斯非线性特性给间歇过程故障监测带来的影响,也考虑了数据的动态特性,提高了间歇过程故障监测的实时性和准确性.数值仿真和工业实例验证了方法的有效性.
Due to the dynamic characteristics of batch process data, the traditional support vector data description (SVDD) method is difficult to realize real-time on-line condition monitoring, and a sliding window-based SVDD online real-time fault monitoring method is proposed. The appropriate size of the sliding window to update the current sub-data space, the establishment of SVDD sub-model, in order to achieve online real-time fault monitoring.This method not only overcomes the non-Gaussian non-linear process data to the intermittent process fault monitoring impact, but also consider the data Which improves the real-time performance and accuracy of intermittent process fault monitoring.The numerical simulation and industrial examples verify the effectiveness of the method.