变压器油中溶解气体色谱峰识别方法的研究及应用
    点此下载全文
引用本文:李永征,宗起振,包素丽.变压器油中溶解气体色谱峰识别方法的研究及应用[J].计算技术与自动化,2020,(2):78-81
摘要点击次数: 927
全文下载次数: 0
作者单位
李永征,宗起振,包素丽 (南京国电南自轨道交通工程有限公司江苏 南京 210032) 
中文摘要:针对目前变压器油中溶解气体分离后得到的色谱曲线基线存在抖动、漂移、有异常点,特别是在低浓度曲线干扰大、波形不规则,易出现对气体色谱峰的误判和漏判,导致气体浓度值检测偏差大,严重影响了故障诊断的准确性等一系列问题。采用了差分阈值滤波法、粗糙惩罚法以及粒子群优化支持向量机等算法,对油气分离后的色谱曲线序列数据中的一系列色谱峰进行干扰峰识别、分类,并采用优化半峰宽法计算出气体浓度值。有效降低了色谱峰的漏判和误判率,极大提高检测气体浓度值的稳定性、准确性和精确性。
中文关键词:油中溶解气体色谱峰  支持向量机  粗糙惩罚法  阈值滤波
 
Research and Application of Chromatographic Peak Identification Method for Dissolved Gas in Transformer Oil
Abstract:In view of the separation of dissolved gas in transformer oil chromatographic curve of jitter,baseline drift,have abnormal points,especially in low concentration curve interference,waveform is irregular,the misjudgement of gas chromatographic peak occurring and false negative,leads to the gas density detection deviation is big,seriously affected the accuracy of fault diagnosis and so on a series of problems. Differential threshold filtering method,rough penalty method,particle swarm optimization support vector machine and other algorithms are used to identify and classify a series of chromatographic peaks in the chromatographic curve sequence data after oil and gas separation,and the optimized half-peak width method is used to calculate the gas concentration value. It can effectively reduce the misjudgment and misjudgment rate of chromatographic peak,and greatly improve the stability,accuracy and accuracy of detection gas concentration value.
keywords:oil dissolved gas chromatographic peak  SVM  Roughness penalty method  threshold filtering
查看全文   查看/发表评论   下载pdf阅读器