Research On Partial Discharge Pattern Recognition In Transformer Based On Improved Fuzzy Clustering Algorithm

In the power network of modern power system,power transformer plays an important role in maintaining the safe and stable power supply demand,so the safe operation of power transformer is related to the power reliability of power grid system.However,due to the influence of various factors,the transformer will have internal insulation fault,and the partial discharge of the transformer is closely related to it How to monitor the partial discharge signal of the transformer and take precautions before the fault is very important for the safe and stable operation of the transformer.Therefore,the pattern recognition of partial discharge is very important for judging the fault type of the transformer and identifying the insulation defects.In this paper,according to the order of data extraction,feature extraction and signal recognition of transformer discharge signal,the partial discharge in transformer is studied.In this paper,four kinds of common insulation defects are designed to discharge along the surface,needle plate(contact and non contact)and suspension.In the laboratory,pulse current method and ultrasonic method are used to extract and record the partial discharge signal of transformer?Based on the different types of data generated by different sensors,in order to obtain accurate and reliable features,pulse current method data is transformed into gray-scale image by prpd spectrum to extract typical moment feature parameters for pattern recognition;ultrasonic signal is denoised by improved threshold wavelet denoising,which is undistorted to the greatest extent in signal restoration,and statistical feature parameters are used for feature extraction Data and wavelet packet energy parameters are combined to extract the feature parameters.Finally,in the data processing,principal component dimension reduction is used to optimize the data to reduce redundancy.In the past,the fuzzy c-means algorithm(FCM)was used to recognize the PD pattern.Considering the sensitivity of the initial value of FCM,if there is individual deviation interference in many samples,it is easy to get the local minimum value,so the solution obtained is not the global optimal solution.From this point of view,it affects the recognition rate in the PD pattern recognition.AFSA is an intelligent optimization algorithm based on natural biological behavior.Its remarkable advantage is that it has powerful global search ability and full range search function,which overcomes the shortage of FCM falling into local optimal solution,and the convergence speed of the algorithm is very fast.So in this paper,the traditional fuzzy c-means clustering algorithm is improved by using artificial fish swarm algorithm(AFSA)In the pattern recognition of partial discharge of transformer,the success rate of recognition is improved.In this paper,the problem that the number of iterations increases obviously after the introduction of artificial fish swarm is optimized,and the step length of artificial fish swarm algorithm is optimized,and the correction factor ?is introduced to solve it,so as to reduce the number of iterations without loss of accuracy.Finally,the effectiveness of the algorithm is verified by experiments.In this paper,through the comparative study of pulse current method and ultrasonic method,it is proved that the statistical characteristic parameters and wavelet packet energy parameters have high accuracy in the study of PD pattern recognition of transformer,and the value of fuzzy clustering algorithm improved by artificial fish swarm algorithm in the classification of PD types of transformer,and the improved algorithm in improving the discharge recognition It has obvious effect on other rate.