Pattern Recognition And Feature Analysis Of Partial Discharge UHF Signal Based On Particle Swarm Optimization Wavelet Neural Network

The transformer is the "Heart" of power system and it plays a core role in power conversion,which is related to the reliable power consumption of industry and residents.It is necessary to study the common faults and solutions of the transformer in order to ensure the continuous supply of electricity.Partial discharge is one of the most common faults in the operation of power transformer.It is easy to erode the insulation medium,resulting in insulation deterioration and reducing the insulation strength.What's more serious is that when the insulation medium breaks down,it will cause huge personal equipment safety accidents and bring high economic losses to the country and enterprises.At present,there are many research methods for partial discharge of transformer,but each method has its own advantages and disadvantages.In this paper,various methods were compared and analyzed,and the corresponding research was carried out.First of all,the partial discharge definition and basic theory in power transformer were described.Then,the existing PD detection methods,PD feature signal extraction methods and discharge pattern recognition algorithm were briefly introduced.In this paper,the wavelet analysis method was utilized to extract the characteristic signals of partial discharge,and symN wavelet was used as the basis function,the optimal noise reduction algorithm was selected.Based on the commonly used algorithm formula for extracting feature parameters of partial discharge,the scale parameter and the displacement parameter were adjusted constantly,and then the feature parameters were extracted and the dimensions were reduced by correlation analysis.Based on the extracted effective feature signals,this paper used the wavelet neural network algorithm to identify four common PD patterns:suspended discharge,needle plate discharge,surface discharge and air gap discharge.Subsequently,the algorithm above was optimized,and the partial discharge signals of transformers of Heze Power Grid were used as examples to verify the recognition effect.The results of the case show that the recognition effect is better than before.Based on the above research,a UHF on-line partial discharge monitoring system was established and applied to Heze Power Grid.The system has successfully detected partial discharge signals of transformers and realized the recognition of discharge types and improves the operation reliability of transformers in Heze area.