Based on vibration signal of high voltage circuit breaker,a new method of intelligent fault diagnosis that wavelet packet extracts energy entropy which are used as characteristic vector of the support vector machine(SVM)to construct classifier for fault diagnosis is presented.The acceleration sensors are applied to collecting the vibration data of different states of high voltage circuit breakers based on self-made experimental platform in this method.The wavelet packet are fully applied to analyze the vibration signal and decompose vibration signal into three layers,and wavelet packet energy entropy of each frequency band are as the characteristic vector of circuit breaker failure mode.Then the intelligent diagnosis network is established on the basis of the support vector machine theory.It is verified that the method has a better capability of classification and a higher accuracy compared with the traditional neural network diagnosis method through distinguishing the three fault modes which are tripping device stuck,the vacuum arcing chamber fixed bolt looseness and too much friction force of the transmission mechanism of circuit breaker in this paper.
YANG ZhuangzhuangLIU YangZHOU GuomingLIN XinJI TianLI Bin
In the computational process of very fast transient overvoltage (VFTO), it is essential to find an accurate model for a gas insulated substation. The arcing model of the disconnector is particularly important. The general arcing model is not able to give a good description of the arc development process. In this paper, based on the physical process of arcing and existing arc models (the exponential time-varying resistance model and the segmental arcing models), a dynamic arcing model is proposed, which is divided into two stages before and after the zero crossing. The dynamic arcing model combines hyperbola time-varying resistance and the Mayr model to describe the dynamic process of arcing. The present paper creates an arc model blockset upon the Matlab/Simulink software platform. Moreover for a specific 1100 kV station, VFTO is simulated in detail based on different arcing models. It is demonstrated that the dynamic arcing model can describe the physical arc process precisely and is useful for improving the accuracy of VFTO simulations.
In order to avoid single-phase adaptive reclosure overlap in a permanent fault,accurate identification of the fault types on transmission line is necessary.We present a fault nature identification method for ultra-high voltage alternating current(UHVAC) transmission lines with shunt reactor.The voltage amplitude ratios of the neutral small reactance voltage to terminal voltage under transient fault and permanent fault are calculated.The significant differences of the ratio under the two faults have been analyzed.It is found that the ratio can be a criterion to distinguish the fault type,transient or permanent,accurately.Additionally,we also proposed a method that delays a beat frequency oscillation cycle to decide the fault types for the existence of beat frequency oscillation which will cause misidentification on transient faults.Abundant simulation results of ATP/EMTP show that the voltage amplitude ratio under transient fault is between 0.6 and 0.7,which is bigger than 2.5,the ratio under permanent fault.
A design of data monitoring recorder is introduced to ensure the stable work of circuit breakers.The recorder is mainly adopted to collect data of different mechanical properties of circuit breakers for real-time online working in power-industrial labs,and to set up a database including both correct information and fault information for assessing and diagnosing power systems.The recorder uses the master-slave detection core design which adopts both digital signal processor(DSP) and peripheral interface controller(PIC).Not-AND(NAND) flash memory is adopted as a mass storage unit,and the designed clock is used to identify the data being accessed.Access to data and communications between the upper computer and the lower computer are achieved through a universal serial bus(USB) communication.Results of an experiment further illustrate the accuracy and reliability of the data obtained by the proposed recorder.
XU JianyuanYANG ZhuangzhuangLIN XiangbinLI BinLIU Yang