摘要
通过对开关设备的运行数据进行持续收集和分析,可以预测设备的健康状况和潜在故障风险,为设备的预防性维护提供科学依据。提出基于声音信号图谱特征的变电站开关故障监测方法,首先,通过阵列麦克风采集变电站开关声信号,并通过多径分集接收的方式提高信号整体质量;其次,在主成分分析的基础上生成声信号的声谱图;最后,结合CNN与SVM建立故障监测模型,将声谱图输入上述模型内,采用CNN提取声谱图特征,利用SVM对特征分类,实现变电站开关故障监测。仿真结果表明,所提方法在信号采集与故障监测方面均表现出良好的性能与精度。
Through continuous collection and analysis of operating data of switchgear,the health status and potential failure risk of equipment can be predicted,providing a scientific basis for preventive maintenance of equipment.A substation switch fault monitoring method based on the characteristics of the sound signal spectrum is proposed.First,the sound signal of the substation switch is collected by the array microphone,and the overall signal quality is improved by multi-path diversity reception;Secondly,the spectrogram of acoustic signal is generated on the basis of principal component analysis;Finally,the fault monitoring model is established by combining CNN and SVM,and the spectrogram is input into the model.The features of the spectrogram are extracted by CNN,and the features are classified by SVM to realize the substation switch fault monitoring.The simulation results show that the proposed method has good performance and accuracy in signal acquisition and fault monitoring.
作者
廖华
潘勇斌
申晓杰
袁卫义
LIAO Hua;PAN Yong-bin;SHEN Xiao-jie;YUAN Wei-yi(Nanning Monitoring Center of Ultra-High Voltage Transmission Company of China Southern Power Grid Co.,LTD,Nanning Guangxi 530001,China)
出处
《计算机仿真》
2024年第12期158-161,196,共5页
Computer Simulation
基金
超高压输电公司2022年第二批服务公开招标项目(2022-FW-2-S-ZB2)(二次招标)(招标编号:CG0100022001526725)。
关键词
声信号
声谱图
变电站
主成分分析
故障监测
Acoustic signal
Acoustic spectrogram
Substation
Principal component analysis
Fault monitoring
