基于MTF和ConvNext的电能质量复合扰动识别

Power quality composite disturbance identification based on MTF and ConvNext

针对新型电力系统中电能质量扰动识别时特征冗余、识别精度低、抗噪能力不强的问题,提出一种基于马尔可夫转换场与ConvNext的电能质量扰动分类方法。首先,利用马尔可夫转换场将一维电能质量扰动数据在保留时间相关性的同时,将其转化为二维特征图像;其次,将得到的二维特征图像输入ConvNext进行自动特征提取,从而实现对电能质量扰动的分类。结果表明,该方法有效识别电能质量扰动,并能克服噪声干扰对模型的影响。

Aiming at the problems of feature redundancy,low recognition accuracy,and weak anti-noise ability when identifying power quality disturbances in new power systems,a power quality disturbance classification method based on Markov conversion fields and ConvNext is proposed.First,the Markov transformation field is used to convert the one-dimensional power quality disturbance data into a two-dimensional feature image while retaining the time correlation;secondly,the obtained two-dimensional feature image is in⁃put into ConvNext for automatic feature extraction,thereby achieving Classification of power quality disturbances.The results show that this method effectively identifies power quality disturbances and can overcome the impact of noise interference on the model.