Vibration Anomaly Detection Of Power Transformers Based On Gaussian Mixture Model

Power transformers are important equipment in power systems,and their good operating conditions are the basis for safe and stable operation of power systems.Since the power transformer in operation is affected by various aspects such as load current,operating voltage,and ambient temperature,the vibration analysis of transformers is often not accurate,which is difficult to meet the actual requirements of abnormal detection of power transformers.The operating conditions of power transformers are complex and changeable,this paper divides the operating conditions of power transformers based on the density peak clustering algorithm.For the problem of artificially selecting the clustering center in the density peak clustering,propose an improved algorithm of density peaks clustering that automatically determines the clustering center.The improved algorithm was tested on artificial data sets,and the results show that the improved algorithm is avoiding the influence of artificial factors.According to the surface vibration mechanism of the transformer,the voltage and current of the transformer are selected as characteristic parameters of the operating conditions,and the operating condition dataset is constructed.Divide the operating condition of the transformer with the standard algorithm and the improved algorithm,and the clustering results are compared and analyzed.Considering the division of operating conditions of power transformers,based on Gaussian mixture model analysis and research on abnormal vibration of power transformers.The Hilbert Huang Transform is used to complete the measurement of the transformer surface vibration measured signal representation,and based on the correlation coefficient,the effective components representing the operation conditions of the transformer are selected to construct a power transformer surface vibration data set.According to the improved density peak clustering algorithm divided the operating conditions,and constructed the transformer vibration abnormality detection model considering operation conditions.The vibration anomaly detection simulation is completed on the power transformer surface vibration dataset.