Fault diagnosis of high voltage circuit breaker based on wavelet packet logarithmic energy entropy

A new method based on wavelet packet-logarithmic energy entropy for the fault diagnosis of HV CB (circuit breakers) is presented in this paper. Firstly, the standard signals of normal state and several fault states are decomposed with wavelet packet at 5 layers, and then calculate the logarithmic energy entropy of each node. Secondly, normalize the logarithmic energy entropy value of each node in the various states, and obtains the relative logarithmic energy entropy. And then calculating various fault status each node of the relative logarithmic energy entropy corresponding to the normal state is used as a variation. And use the logarithmic energy entropy´s variation of each node to make-up characteristic curve to reflect the failure status. Finally, characteristic curve which is formed by the testing signals via the same means is compared with standard characteristic curve. And according to the result of comparison we can determine that the signal belongs to which type of fault. After a number of experiments, it show that the method can reflect the various mechanical fault status change of vibration signal of HV CB (circuit breakers) and achieve better diagnostic results.