摘要
为了提高数字化电能计量表故障检测的准确性,研究一种基于神经网络自适应滤波器的数字化电能计量故障检测方法。利用电流传感器、电压传感器以及功率传感器采集三种数字化电能计量装置运行数据并实施去噪和归一化处理。提取数据的时域和频域两类四种特征参数,包括方根幅值、偏斜度、频谱幅值、均方根频率。以特征为输入,神经网络自适应滤波器训练后,实现数字化电能计量故障检测。结果表明:检测结果的Z值均在8以上,说明检测结果更接近所设置的真实故障类型,检测更为准确。
In order to improve the accuracy of digital electric energy meter fault detection,a fault detection method of digital electric energy meter based on neural network adaptive filter is studied.The current sensor,voltage sensor and power sensor are used to collect the operation data of three kinds of digital electric energy metering devices and implement noise elimination and normalization processing.Two types of four characteristic parameters of data are extracted in time domain and frequency domain,including square root amplitude,deviation,spectrum amplitude and root mean square frequency.With the feature as the input,the neural network adaptive filter is trained to realize the fault detection of digital electric energy metering.The results show that under the application of the method in this paper,the Z values of the detection results are all above 8,which indicates that the detection results are closer to the real fault types set,and the detection is more accurate.
作者
张龙
沈飞
李林林
李兴渊
王磊
ZHANG Long;SHEN Fei;LI Lin-lin;LI Xing-yuan;WANG Lei(CHN Energy Shendong Saftey Administra Tration,Erdos 017200 China;CHN Energy Shendong Coal Shangwan Collieery,Erdos 017200 China)
出处
《自动化技术与应用》
2024年第4期47-50,176,共5页
Techniques of Automation and Applications
关键词
神经网络自适应滤波器
数字化电能计量装置
特征提取
故障检测方法
neural network adaptive filter
digital electric energy metering device
feature extraction
fault detection method
