一种变压器色谱基线自适应扣除方法的研究

为了自适应地扣除变压器油中溶解气体色谱基线,提出了基于半峰宽和轮廓提取的基线扣除方法。基于轮廓提取的算法是通过构造一个滑动窗,沿着色谱数据滑动求窗内的最小值,将其插值平滑即可获得色谱基线;基于小波和轮廓提取的算法是先用小波初步提取基线,将其从色谱数据中减去并用轮廓提取算法获得基线,将小波提取的和轮廓提取的基线相加即为原数据的基线。针对这些算法效果与滑动窗宽度有关的缺点,提出了变压器色谱基线自适应扣除方法:用基于小波和轮廓提取的算法扣除基线,由所得数据求取每个成分的半峰宽,以最大的半峰宽乘以一系数为所需滑动窗宽,再以此窗宽和基于小波和轮廓提取的算法获得基线。自适应算法与基于小波和轮廓提取的算法比较实验研究的结果表明,前者比后者效果好,与滑动窗宽度关系不大,能自适应地提取变压器色谱基线。该法能广泛用于变压器油中溶解气体检测领域。

Transformer Chromatogram Baseline Self-adapting Deduction Method

To self-adaptedly deduct the chromatogram baseline of dissolved gas in transformer oil,a half peak width and profile extraction-based baseline deduction method was proposed.The algorithm of the profile extraction-based was as follows,firstly,constructing a sliding window,secondly,sliding the window following the chromatogram data while recording the minimums data of the window,finally,interpolating and smoothing the minimums,then the baseline was obtained.The algorithm of the wavelet and profile extraction-based was as follows,firstly,extracting a preparatory baseline using wavelet filter,secondly,subtracting the preparatory baseline from the chromatogram data,thirdly,extracting baseline from the subtracted data using the profile extraction-based algorithm,finally,adding up the two baselines,the real baseline was formed.To weaken the relationship between the above methods' effect and the width of the sliding widow,a self-adapting algorithm was proposed,namely,deducting baseline from the chromatogram data using the wavelet and profile extraction-based method,then,calculating the half peak width of every component from the deducted data,thirdly,multiplying the maximum half peak width by a factor,finally,taking the multiplied half peak width as the sliding window width and applying the wavelet and profile extraction-based method to the original chromatogram data,then the real baseline was extracted.An experiment was performed to compare the self-adapting method and the results show that self-adapting method is superior to the latter,it is hardly related to the width of the sliding window and can self-adaptedly extract baseline from the transformer chromatogram data.The suggested method can be widely applied to the field of transformer oil dissolved gas measurement.