Research On Circuit Breaker Fault Recognition Based On Multi Information Fusion

High voltage circuit breaker as control and protection devices is very important in the power system, its operation, maintenance and overhaul ofrelationship between the power system can run reliably. According to thesurvey data found that the reliability of circuit breaker CIGRE on which is in use, which occupies in the mechanical fault many fault influence the reliability of high voltage circuit breaker in the proportion reached 70% ~80%, so it is necessary to mechanical characteristics of high voltage circuit breaker monitoring and on which it has produced or potential faultrecognition. At present, mostly based on the online monitoring of a single feature recognition research results of HV circuit breaker mechanical state,and to identify research at home and abroad is still relatively rare on the operation of circuit breaker working state and fault mode from multi featureangle. In view of this, this paper integrated circuit breaker moving contact and speed parameters of vibration signals and acoustic signal switchingprocess, from the perspective of running state analysis of the circuit breaker, and combined with the method of multi information fusion in the rough set theory and support vector machine for circuit breaker operation mode recognition.The running status of first use of high speed and high resolution camera to capture the switching process of transmission parts of the circuit breaker,using TLD video tracking technique to obtain the circuit breakertransmission compon ents in displacement process and parametercalculation of switching, maximum ve locity and average velocity.Then using the ensemble empirical mode decomposition(EEMD) method of vibration signal and acoustic signal closing process the collected weredecompos ed into a set of intrinsic mode function(IMF), and combined with the theory of information entropy, extract the energy entropy of eachintrinsic mode functi on as the signal characteristic quantity. Because thevibration signals and acousti c signal after the decomposition will get more IMF energy entropy that high di mension feature, in order to reduce the dimensionality of feature dimension an d vibration signals and acousticsignal characteristic quantity of the best use of i nformation fusion theory inrough set attribute reduction of vibration signals an d acoustic signal characteristic quantity.Finally, using support vector machine(SVM) mode of operation of circuit breaker for the identification, pattern recognition effect and the comparison of the experimental data of concrete analysis using a single characteristic quantity of the signals and multi information fusion mode recognition effect,the experimental results show that the fault pattern recognition methodbased on multi information fusion has a good effect.