During the opening and closing operations of high-voltage circuit breakers, vibration signals are generated that contain information such as the movement trajectory, contact state, and wear degree of mechanical components of the equipment. To address the problems such as incomplete information and indistinct features when single-channel vibration signals reflect the mechanical state of high-voltage circuit breakers, a feature processing method that fuses the voiceprint attributes of multi-channel vibration signals is proposed. Firstly, multi-channel vibration signals of high-voltage circuit breakers are collected and processed by bispectrum analysis. According to the symmetry and self-similarity of the bispectrum, the bispectrum spectrograms of each channel are segmented under the premise of ensuring the integrity of information; Then, the brightness factor is proposed according to the vibration intensity distribution of each channel. It is used to perform weighted fusion on the bispectrum spectrum to achieve the convergence of voiceprint attributes of different channels into the same spectrum. The fused spectrogram is obtained; Finally, the color and texture features of the fused spectrum are extracted. They are optimized by factor analysis to obtain voiceprint attribute fusion (VAF) features. Experiments show that this method realizes multi-channel signal weighted fusion and improves the comprehensiveness and integrity of voiceprint features. It has the ability to describe equipment status information in an all-round and multi-angle way. It provides a new method for vibration feature extraction of high voltage circuit breaker.