摘要
提出了一种基于短时傅里叶变换和DAG(Directed Acyclic Graph)支持向量机的电能质量扰动检测和识别方法。将扰动信号通过Blackman窗短时傅里叶变换,得出时域最大幅值向量,然后把它作为特征向量输入到DAG支持向量机来实现电能质量扰动类型的识别。其中,时域最大幅值向量不仅能反映各种扰动的特征,还能显示电压突升、电压暂降、电压中断和暂态振荡等扰动的发生时刻和持续时间。仿真测试表明,该方法能有效识别各种电能质量扰动,而且识别正确率高,训练时间短,实时性能较好。
The paper proposes a new method for power quality(PQ)disturbances detection and identification based on short time Fourier transform and directed acyclic graph support vector machines(DAGSVMs).The time-domain maximum amplitude vector can be obtained by Blackman window short time Fourier transform of disturbance signals.Then the method makes it as the eigenvector and sets it to the DAGSVMs to realize the identification of the power quality disturbance type.Besides,the time-domain maximum amplitude vector not only can reflect the character of each disturbance,but also can show the starting and sustainable time of voltage swell,voltage sag,interruption and oscillatory transients.Simulation results show that the method could detect and classify the PQ disturbance effectively,and the classifier has a good performance on correct ratio,training speed and real-time function.
出处
《电力系统保护与控制》
EI
CSCD
北大核心
2011年第1期83-86,103,共5页
Power System Protection and Control
关键词
短时傅里叶变换
Blackman窗
特征提取
支持向量机
DAG
short time Fourier transform
Blackman window
feature extraction
support vector machines
directed acyclic graph