摘要
针对暂态电能质量电压多扰动信号的检测与分类问题,提出一种基于广义S变换及模糊SOM神经网络的暂态电能质量检测和识别方法。针对常见的电压多扰动信号,特别是两种扰动叠加的情况,采用广义S变换对扰动信号的时频特征进行提取,并取变换后的时间幅值平方和均值和特征频点作为神经网络的输入样本,采用模糊SOM神经网络进行训练,再用新的多扰动数据进行网络检验。仿真与实验结果表明,广义S变换能有效提高电能质量多扰动特征检测,模糊SOM神经网络能精确对其进行分类,该方法能够较好的解决电压多扰动叠加情况的定性和定量分类问题。
To solve the problem of detecting and classifying power quality multi-disturbances, this paper proposed a new method based on the generalized S-transform and the fuzzy self-organizing maps（SOM） neural network to extract features and to recognize the disturbance patterns. As to all kinds of the disturbance voltage signals, especially the superposition of two kinds of voltage disturbances, the generalized S-transform is used to extract multi-disturbance time-frequency features. Then, the average square-sum of S-transform amplitudes are used to train the fuzzy SOM neural network, and the new collected data are tested using the trained fuzzy SOM neural network. Simulation and experiment results show that the generalized S-transform can detect power quality multi-disturbance effectively, and the fuzzy SOM neural network can classify it accurately. The problem of voltage super imposed disturbance classification can be resolved successfully from both qualitative and quantitative ways.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2015年第4期866-872,共7页
Proceedings of the CSEE
基金
国家杰出青年科学基金项目(50925727)
国家自然科学基金项目(60876022)
国防预研重大项目(C1120110004)~~
关键词
电能质量检测
多扰动检测
S变换
广义S变换
SOM神经网络
power quality detection
multi-disturbance detection
S-transform
generalized S-transform
SOM neural network
