【摘要】 针对回转窑过程多时间尺度信号的耦合关系往往会掩盖过程变化规律,导致未经时间尺度分解的直接的因子分析方法难以得到过程固有的变化模式的问题,提出采用基于EMD的多尺度因子分析方法对锌钡白煅烧过程的数据进行分析。该方法利用EMD有较强的自适应性,分解结果往往具有较强的物理意义等特性,首先对锌钡白回转窑平稳运行时的生产数据进行EMD分解,根据分解结果提取不同的时间尺度数值,然后对不同尺度下的数据进行因子分析。分析结果显示,数据在周期3 h以上的大时间尺度和周期30 min以下的中小时间尺度上都分别表现出定常的特点,表明人工控制可分解为长周期控制和短周期控制两方面。实验结果表明该方法在处理回转窑过程多尺度数据处理方面的有效性。更多还原

【Abstract】 The coupling of variables in rotary kiln processes under multiple time-scale usually conceals the underlying principles,which results in the difficulty in revealing the inherent mode of the process directly using the factor analysis.An improved factor analysis method based on EMD is used to deal with the data of a lithopone calcination process.EMD is a self-adaptive method of signal decomposition,which leads to a practically result.The field data is decomposed by EMD method,based on that several different time-scale values are chosen.The dataset under different scales are analyzed by factor analysis method.The analysis result shows that the data are steady-state both under large scale(more than 3 hours) and small scale(less than 30 minutes).It can be inferred that manual control scheme should be divided into a short-term scheme and a long-term one.This experiment proves the effectiveness of the proposed method when applying to multi-scale data processing of rotary kilns.更多还原