遗传算法优化GMM的故障电弧识别方法
|
摘要
由于电气设备中正常电流信号和故障电流信号电流特性十分相似,故障电弧的正确识别十分困难,但非常重要。采用小波变换结合近似熵的方法对故障和正常电流特征进行提取,将特征用来训练混合高斯模型(GMM)。但GMM的参数估计方法期望最大(EM),极易局部收敛,因此,引进遗传算法,提出遗传优化的高斯混合模型(GA-GMM)。用经过特征提取后的特征数据来训练GA-GMM模型,模型根据其输入的数据特征得到最大的概率分类,即输出电流故障与否。大量实验表明采用GA-BMM能够快速准确地对故障电弧特征进行识别,达到了较好的识别效果,实现了准确识别故障电弧的操作。
Since the fact that normal and fault current characteristics are similar in electrical equipment, it is very difficult and important to accurately identify the faulty arcs. The wavelet transform combined with the approximate entropy is used to extract the fault and normal current characteristics, which are used for the training of the Gaussian mixture model(GMM). However, because of GMM parameters estimation method of expectation maximum(EM), local coverges easily, the genetic algorithm was introduced, and a genetic optimized GMM(GA-GMM) was proposed. The GA-GMM model is trained by the feature extraction data. The maximum probability classification is obtained according to the input data features, that is, the output current is normal or faulty. A large number of experiments show that GA-GMM can quickly and accurately identify the characteristics of the fault arc, achieve a better recognition effect, has realized the accurate identification of arc fault operation.
Since the fact that normal and fault current characteristics are similar in electrical equipment, it is very difficult and important to accurately identify the faulty arcs. The wavelet transform combined with the approximate entropy is used to extract the fault and normal current characteristics, which are used for the training of the Gaussian mixture model(GMM). However, because of GMM parameters estimation method of expectation maximum(EM), local coverges easily, the genetic algorithm was introduced, and a genetic optimized GMM(GA-GMM) was proposed. The GA-GMM model is trained by the feature extraction data. The maximum probability classification is obtained according to the input data features, that is, the output current is normal or faulty. A large number of experiments show that GA-GMM can quickly and accurately identify the characteristics of the fault arc, achieve a better recognition effect, has realized the accurate identification of arc fault operation.
引文
[1]公安部消防局. 2016年全国火灾情况分析[EB/OL]. http://www. 119. gov. cn/xiaofang/nbnj/34601. htm2017.
[2]张士文,张峰,王子骏,等.一种基于小波变换能量与神经网络结合的串联型故障电弧辨识方法[J].电工技术学报,2014,29(6):290-295.
[3]王子骏,张峰,张士文,等.基于支持向量机的低压串联故障电弧识别方法研究[J].电测与仪表,2013(4):22-26.
[4]韩旭,王蒙.基于遗传算法优化BP神经网络的故障电弧识别[J].测控技术,2016,35(12):21-25.
[5]陈照,李奎,张洋子,等.基于电弧电磁辐射的故障电弧识别[J].电工电能新技术,2017,36(3):70-74.
[6] Liu Y W,Wu C J,Wang Y C. Detection of serial arc fault on low-voltage indoor power lines by using radial basis function neural network[J]. International Journal of Electrical Power and Energy Systems,2016,83:149-157.
[7] Wang J G,Xu C,Sun J X. Experimental study of fault arc protection based on UV pulse method in high voltage switchgear[J]. International Journal of Emerging Electric Power Systems,2014,34(10):99-102.
[8]刘晓明,赵洋,曹云东,等.基于小波变换的交流系统串联电弧故障诊断[J].电工技术学报,2014,29(1):10-17.
[9]郭凤仪,李坤,陈昌垦,等.基于小波近似熵的串联电弧故障识别方法[J].电工技术学报,2016,31(24):164-172.
[10]宋东奇.基于高斯混合模型的基因表达数据聚类研究[D].镇江:江苏大学,2015.
[11]赵梦玲.基于智能优化算法的聚类分析及应用[D].西安:西安电子科技大学,2015.
[12]马永杰,云文霞.遗传算法研究进展[J].计算机应用研究,2012,29(4):1201-1206.
[13]骆俊,马尽文.高斯混合模型的遗传分基融合算法[C]//第十二届全国信号处理学术会议.2005.
[14]钟金琴,辜丽川,檀结庆,等.基于EM算法的GMM参数估计[J].计算机工程与应用,2012,48(34):28-32.
[15]张永鑫,冉鑫.基于在线合并EM算法的高斯混合模型分类方法[J].计算机应研究,2010,27(8):2906-2908.
[16]谢娟英,张琰,王春霞,等.基于遗传算子的遗传算法和自适应遗传算法[J].计算机工程与应用,2010,46(33):28-31.
[2]张士文,张峰,王子骏,等.一种基于小波变换能量与神经网络结合的串联型故障电弧辨识方法[J].电工技术学报,2014,29(6):290-295.
[3]王子骏,张峰,张士文,等.基于支持向量机的低压串联故障电弧识别方法研究[J].电测与仪表,2013(4):22-26.
[4]韩旭,王蒙.基于遗传算法优化BP神经网络的故障电弧识别[J].测控技术,2016,35(12):21-25.
[5]陈照,李奎,张洋子,等.基于电弧电磁辐射的故障电弧识别[J].电工电能新技术,2017,36(3):70-74.
[6] Liu Y W,Wu C J,Wang Y C. Detection of serial arc fault on low-voltage indoor power lines by using radial basis function neural network[J]. International Journal of Electrical Power and Energy Systems,2016,83:149-157.
[7] Wang J G,Xu C,Sun J X. Experimental study of fault arc protection based on UV pulse method in high voltage switchgear[J]. International Journal of Emerging Electric Power Systems,2014,34(10):99-102.
[8]刘晓明,赵洋,曹云东,等.基于小波变换的交流系统串联电弧故障诊断[J].电工技术学报,2014,29(1):10-17.
[9]郭凤仪,李坤,陈昌垦,等.基于小波近似熵的串联电弧故障识别方法[J].电工技术学报,2016,31(24):164-172.
[10]宋东奇.基于高斯混合模型的基因表达数据聚类研究[D].镇江:江苏大学,2015.
[11]赵梦玲.基于智能优化算法的聚类分析及应用[D].西安:西安电子科技大学,2015.
[12]马永杰,云文霞.遗传算法研究进展[J].计算机应用研究,2012,29(4):1201-1206.
[13]骆俊,马尽文.高斯混合模型的遗传分基融合算法[C]//第十二届全国信号处理学术会议.2005.
[14]钟金琴,辜丽川,檀结庆,等.基于EM算法的GMM参数估计[J].计算机工程与应用,2012,48(34):28-32.
[15]张永鑫,冉鑫.基于在线合并EM算法的高斯混合模型分类方法[J].计算机应研究,2010,27(8):2906-2908.
[16]谢娟英,张琰,王春霞,等.基于遗传算子的遗传算法和自适应遗传算法[J].计算机工程与应用,2010,46(33):28-31.