摘要
本文提出一种基于振动信号频谱高斯混合模型(GMM)的变电站瓷支柱绝缘子振动信号特征提取方法,并利用粒子群算法(PSO)优化后的极限学习机(ELM)实现故障状态识别与分类。首先,根据瓷支柱绝缘子振动信号的频谱得到信号的频域分布。然后,利用高斯概率密度函数和期望最大值算法(EM)将频谱划分为三种模态,每个模态可以得到标准差σ、权重系数α和均值μ三个特征参数用于表征各频率的模态带宽、模态分量占比和模态中心频率。最后,将各模态的特征参数作为特征值输入分类模型实现状态识别与分类。
A vibration signal feature extraction method based on Gaussian mixture model(GMM)of vibration signal spectrum is proposed,and the extreme learning machine(ELM)optimized by particle swarm optimization(PSO)is used to realize fault state recognition and classification.Firstly,the frequency spectrum of the vibration signal of porcelain post insulator is obtained.Then,Gaussian probability density function and expectation maximization algorithm(EM)are used to divide the frequency spectrum into three modes.Three characteristic parameters,standard deviationσ,weight coefficientαand meanμ,can be obtained for each mode to characterize the modal bandwidth,modal component proportion and modal center frequency.Finally,the characteristic parameters of each mode are input into the classification model as eigen values to realize state recognition and classification.
作者
焦宗寒
邵鑫明
郑欣
刘荣海
JIAO Zonghan;SHAO Xinming;ZHENG Xin;LIU Ronghai(Electric Power Research Institute of Yunnan Power Grid Co.,Ltd,Kunming 650217;Graduate Workstation of Yunnan Power Grid Company,North China Electric Power University,Kunming 650217;Hebei Key Laboratory of Electric Machinery Health Maintenance&Failure Prevention,North China Electric Power University,Baoding,Hebei 071003)
出处
《电气技术》
2021年第6期36-42,共7页
Electrical Engineering
基金
云南电网内部科技项目(YNKJXM20180729)。
关键词
瓷支柱绝缘子
振动信号频谱
高斯混合模型
故障诊断
porcelain post insulator
vibration signal spectrum
Gaussian mixture model(GMM)
fault diagnosis
