Mechanical Fault Diagnosis For High Voltage Circuit Breakers Based On Vibration Signals

Circuit breaker action generates vibration signal can reflect the mechanical properties ofcircuit breaker, using the information processing technology to the vibration signal of themechanical failure and collecting useful characteristic fault diagnosis is the hot point of theresearch of fault diagnosis, and because of its noninvasive collection way, makes it morelikely to collect accurate data, the application foreground is very wide. Based on vibrationsignal, wavelet packet and relevance vector machine principle applied to high voltage circuitbreaker mechanical fault diagnosis is a research.In order to obtain fault samples, this paper use ZW-32permanent magnetic actuatortype vacuum circuit breaker as the research object, points brake spring falls off, permanentmagnetic actuator screw loosing and excitation current is too large cases of vibration signalacquisition, used for failure analysis.As the core part, this paper focuses on extracting the signal eigenvectors of vibrationsignal by wavelet packet and using the principle of relevance vector machine diagnosefailures. Researches show that, circuit breaker failure vibration signal characteristic value canbe well extracted by the wavelet packet, clustering is very well. On that basis, Thedissertation takes as samples eigenvalue extracted by wavelet packet,uses the principle ofrelevance vector machine diagnose failures model. Through the simulation found that thismethod is better than the traditional neural network, vector machine theory fault diagnosismodel has higher accuracy and faster speed etc.Finally, the application of the above method, the establishment of circuit breaker faultdiagnosis model for the collected samples failure analysis. Test results show that this methodhas a higher fault recognition ability, so it has great practical value.