Recognition of Complex Abnormal Vibration Voiceprint of Power Transmission and Transformation Equipment in Cold Environment

The transmission equipment is a vibration signal measurement system. The vibration characteristics under different mechanical and electrical fault conditions are analyzed to determine the relationship between on-load tap changes (OLTC) conditions and vibration signals. However, in low- temperature environments, the embrittlement of equipment materials and solidification of lubricants occur and change the vibration signals. This increases the complexity of identifying abnormal vibration voiceprints. Therefore, we developed a voiceprint recognition method for complex abnormal vibrations of transmission equipment in cold environments. Combining Empirical Mode Decomposition (EMD) with Stein’s unbiased estimate (SURE), the vibration of the equipment was measured. SURE was used to collect vibration voiceprint signals of power transmission and transformation and denoise them. The denoised voiceprint signal was divided into multiple voiceprint segments and spectrograms. In feature extraction, the normal spectrogram was used as input, while the normal spectrogram was used as input. A Long Short Term Memory (LSTM) network was used for training to classify the input spectrogram voiceprint samples and abnormal samples for the accurate recognition of abnormal vibration of the voiceprint of power transmission and transformation equipment. The experimental results showed that the proposed method accurately identified the abnormal vibration voiceprints of power transmission and transformation equipment. The developed method identified the abnormal vibration voiceprints of power transmission and transformation with a recognition rate of over 96% and a recognition time of 94.47 ms.