High voltage circuit breaker generates multiple signals during its operation

which plays an indispensable role in diagnosing its operation state. The traditional fault diagnosis ideas of high-voltage circuit breaker mostly start from the single sensor acquisition signal for analysis

and apply typical classification algorithms for fault signal identification. However

the robustness of the single type of sensor is not strong

which is prone to cause the problem of signal acquisition integrity due to the shortcomings of the sensor itself

and the subsequent classification algorithms do not perform selection optimization for the characteristics of the timing signal. In order to solve above-mentioned problems

in this paper a multi signal joint intelligent fault diagnosis method for high-voltage circuit breaker based on VMD-BiLSTM is proposed

the sound and vibration sensors at different installation locations are used for signal acquisition. Moreover

the traditional one-dimensional VMD algorithm is improved so to greatly reduce the signal processing time under the same working condition. Finally

the decomposition matrix is extracted with such two time-frequency domain features as instantaneous frequency and spectral entropy. Three kinds of time series classification algorithms for industrial small sample data and one kind of traditional unordered image algorithm are used for fault classification and diagnosis of typical working condition feature data set. The experimental results show that the diagnostic accuracy of the fault signal recognition model described in this paper can reach 92.4%under the main seven working conditions

which has a good application prospect.