Mechanical Fault Diagnosis Of On-load Tap-changer Based On Vibration Analysis

On-Load Tap-Changer(OLTC)can change the voltage ratio with load and undertake important functions such as adjusting system voltage and power,compensating voltage fluctuations and improving power quality under uninterrupted power conditions.It is an important component to ensure the safe and stable supply of electricity.In the event of a fault,the component can be damaged.With the further development of the fault,the transformer will be burnt down and the quality of the power supply is degraded,resulting in a shortage of power supply in the area.Studies have shown that the main fault of OLTC is mechanical failure.Considering that the vibration signal during OLTC switching process is closely related to its mechanical condition,and the non-intrusive acquisition of the vibration signal does not affect the normal operation of the transformer,the OLTC mechanical condition monitoring and diagnosis based on the vibration signal has received a lot of attention from domestic and foreign researchers.However,the mechanical structure of OLTC is complex and there are many types of faults.It is thus necessary to carry out OLTC mechanical fault diagnosis based on vibration signals to improve its operational reliability.The main tasks are listed as follows:Based on the summary of OLTC fault types,a typical mechanical fault test scheme for a combined OLTC is designed.The vibration signals under normal and typical mechanical fault conditions are tested,which provides important data support for the subsequent study.Considering the function of OLTC components and their influences on the vibration signals,Tunable Q-factor Wavelet Transform(TQWT)optimized by an artificial fish swarm algorithm is proposed based on the energy concentration of the OLTC vibration signal.The vibration signal decomposition method quantitatively describes the energy of the OLTC vibration signal subsequence obtained by the decomposition using the grey correlation degree.The results show that the optimized TQWT decomposition method improves the accuracy of OLTC vibration signal decomposition results effectively.The grey correlation indexes of the vibration signals in the normal condition are above 0.9,indicating the consistency of vibration signals as well as the agreement of characteristics during the OLTC switching process.However,the proposed grey correlation index has limited ability to distinguish different fault types of OLTC,which has to cooperate with the study of pattern recognition.To further improve the accuracy of OLTC typical mechanical fault diagnosis,a typical mechanical fault identification method based on Hidden Markov Model(HMM)is studied.Based on the OLTC vibration signal subsequence energy,a HMM library is established for signals of different measuring points,fault types and different level of the fault.According to the library,signals are analyzed and judged on considering the corresponding types and levels of the fault.The results show that the confidence ratio of the diagnosis under all conditions can be maintained above 95%,which indicates that the HMM library combining the extracted feature quantities can effectively identify various faults and distinguish their fault development stages,which has great engineering significance.The research results can provide a significant reference for the data analysis,monitoring,mechanical condition diagnosis and the proposal of maintenance strategy of OLTC.