无人值守变电站电力设备音频监测及故障诊断系统

针对智能变电站的无人值守需求及现有故障诊断系统的不足,提出一种电力设备音频监测及故障诊断系统。根据变电站电力设备音频信号信噪比较低的特点,采用具有强鲁棒性的梅尔频率倒谱系数作为判断音频信号异常的特征参数,在此基础上根据音频特征构成多样本观测序列,并采用隐马尔科夫模型进行故障诊断,通过对比对数似然估计概率的输出值确定故障类型。该方法具有实时性较强的优势,也避免了现有故障诊断方法要求较大样本容量的缺陷。实验结果表明,该故障诊断系统具有较高的识别率和鲁棒性。

Fault Diagnosis and Condition Monitoring of Smart Substation Equipment based on Acoustic Signals

Aiming at the requirement of unattended intelligent substation and the shortcomings of existing fault diagnosis system, an audio monitoring and fault diagnosis system of power equipment is proposed. According to the low signal-to-noise ratio (SNR) of the audio signal in substation power equipment, the Mel-frequency cepstrum coefficients with strong robustness are used as the characteristic parameters to judge the audio signal anomaly. Based on the audio feature, a multi-sample observation sequence is constructed. The hidden Markov model (HMM) is used to diagnose the fault, and the fault type is identified by comparing the logarithm likelihood estimate output value. The method has the advantage of real-time and avoids the limitation of the existing fault diagnosis method which requires large sample size. The experimental result shows that the proposed fault diagnosis system has high recognition rate and robustness.