Research On Condition Monitoring And Fault Diagnosis Of Power Transformer Based On Vibration Signal Analysis

Vibration analysis method is the technique to diagnose the conditions of power transformer by measuring and analyzing the vibration signals of its tank, and it is a good supplement for traditional methods based on electrical measurement. In this paper, vibration analysis method is adopted and transformer's core and winding are taken as research objects, their vibration characteristics under normal and unhealthy status are studies by theoretical analysis combined with experimental demonstration. The researches refer to the following aspects:the force analysis for core and winding, the influence on their vibration performances because of winding deformation and core compression condition, how to accurately collect the vibration signals from the surface of power transformer tank, the selection of optimal position of vibration sensors, the relation between vibration signals and applied voltage as well as load current, how to build up vibration prediction model, the expansibility of vibration model, features denotation and extraction for tank vibration signals, etc.The research is conducted in order to solve the above problems. The main work and conclusions are described as follow:The research status of monitoring and diagnosis methods for power transformer at home and abroad is introduced. The existing shortcomings in vibration analysis method are analyzed emphatically. Features extraction methods for vibration signals need to be corrected and improved. Besides, current vibration prediction models, which are mostly based on finite element method, are complicated to build up accurately.The construction and transmission route are studies firstly. The equivalent mathematical model for the winding vibration is proposed and force analysis is conducted. On the basis of force analysis results, the relation between winding vibration and compression situation is investigated in theory. When the compression is degenerate the winding vibration acceleration will enlarge. On the other hand, the principle of core magnetostriction and its influencing factors are presented.The design process of on-line monitoring platform based on vibration measurement is introduced, such as the selection and fixation of vibration sensors, the design and realization of hardware and software as well as anti-interference treatment. By using the measuring platform, the tank vibration signals are picked up on the surfaces of a single-phase transformer and a three-phase one. The results show that the platform can fulfill the requirements for monitoring the core and winding conditions of power transformer. The research on the optimal position for vibration sensors are carried out on the three-phase transformer. Besides, the changing law of load current at different time is traced.According to the principles of generation and transmission of transformer vibration, the tank vibration prediction model is proposed using mechanical vibration theory. The parameters identification is completed by non-negative linear least square method, and then the vibration model is utilized to predict the tank vibration under different load and applied voltage while the prediction value is proved to coincide with the measured one. The model is adopted to detect the simulation winding deformation fault, and the results show the validity of the model. Considering the theoretical basis and acquisition of input variables the feasibility of expansibility of tank vibration prediction model, including the influence of temperature, on-load tap-changers vibration as well as timeliness of model, is investigated.The application of Wavelet Packets Transform (WPT) to monitor the core compression condition is summarized, while the principal shortages are analyzed and pointed out. Considering the characteristics of core vibration signal a simulation signal is constructed for analyzing the intrinsic shortcomings of Hilbert-Huang Transform (HHT) to process tank vibration signals. In order to make use of the advantages and overcome the deficiencies of WPT and HHT, a time-scale-frequency analysis method is developed based on the combination of these two techniques. In this method, the appropriate mother wavelet is established by calculating and quantifying the mean error (ME) and the percent residual difference (PRD) between original signals and reconstructed ones, and then the desirable wavelet packet nodes corresponding to characteristic frequency bands of power transformer are selected through a Correlation Degree Threshold Screening (CDTS) technique, finally HHT is carried out to extract the vibration features in time-frequency domain. The proposed method is used for analyzing and diagnosing the core compression and loosening faults. Compared with the results using the methods in related references, the features obtained by the proposed method reveal clearer condition pattern of the transformer, which shows the potential of this method in condition monitoring of power transformer. This research is sponsored by Natural Science Foundation of Anhui Province (No.070414153).