论文部分内容阅读
趋势能够反映过程重要参数的运行状态和随后的发展趋势。本文采用一种简洁的趋势提取方法,从变量中提取半定量的片段,用3个基元来描述片段:不变、上升、下降,这种方法能够检测过程变量的跃迁,计算复杂度低,基元之间的相关性小,提取的趋势直观、简洁。针对实际工业过程中实时处理的需要,采用一种外推式在线分割算法,实现了对在线数据的合理分割,再结合最小二乘法对在线数据进行分段拟合,使其能满足在线处理的需要。
Trends reflect the operational status of the important parameters of the process and their subsequent trends. In this paper, we adopt a simple trend extraction method to extract semi-quantitative fragments from variables and describe the fragments by three primitives: invariant, rising, and descending. This method can detect the transition of process variables with low computational complexity, The correlation between primitives is small, the extraction trend is intuitive and concise. In order to meet the need of real-time processing in real industrial process, an extrapolation-based on-line segmentation algorithm is adopted to realize the reasonable segmentation of online data, and then the least squares method is used to segment the online data to fit the online processing need.