Electronic Transformer Fault Detection System Based On Wavelet Packet Energy Spectrum And Neural Network

The power transformer is an indispensable electrical measuring device between the primary equipment and the secondary equipment in the power system.Whether the transformer can operate accurately and reliably determines whether the power system can operate reliably and safely.At present,the traditional electromagnetic transformer used in the power system,the complex insulation structure,large volume,small dynamic range,and large current,the saturation of CT affects the fault identification of the secondary protection devices.The analog signal output from the transformer cannot directly interface with the metering and protection devices,and produce ferromagnetic resonance,A series of shortcomings such as magnetic resonance have become increasingly unable to adapt to the development of power systems.With the country’s construction of smart grids and ultra-high voltage transmission systems,there is an urgent need for a new type of transformer for electrical measurement of the transmission system.Electronic transformers are born from this,and the electronic transformers process the collected data on-site.The digital signal is directly sent to the secondary equipment,eliminating the need for the secondary equipment to process the data,and the electronic transformer has a great improvement in bandwidth,insulation,cost and maintenance compared with the electromagnetic transformer,and not subject to electromagnetic interference transmitted by ultra-high voltage power grids.The rapid development of electronic technology and computer measurement technology has also greatly promoted the development of electronic transformers.However,whether the electronic transformer can work stably under various conditions as a new thing has become the core issue of whether it can be practical.Therefore,the fault detection of electronic transformers is very important.Based on the analysis of electronic transformers and the research of domestic and foreign fault detection technologies,this paper proposes a joint fault detection algorithm based on wavelet packet transform to extract energy spectrum and neural network.The method combines signal processing technology and knowledge-based processing technology,extracts the fault feature quantity using wavelet packet energy spectrum,and uses neural network to identify the fault type.This article will conduct specific research through the following aspects:1)This article introduces the structure,principle and research status of electronic transformers at home and abroad,and uses simulation experiments to obtain fault models,and uses mathematical models to classify faults,and obtains the mathematical model of transformer faults.2)The calculation method and advantages of wavelet packet transform are introduced in detail,and the method of extracting wavelet packet energy spectrum is proposed.Then the advantages and disadvantages of neural network are introduced and BP neural network is selected for fault classification.3)Perform fault simulation and data analysis on electronic transformers based on the above theory,and send the processed experimental data to the trained neural network.The three types of verification results show that the fault detection method based on wavelet packet energy spectrum and neural network is correct and a single transformer has a better detection result.4)A joint diagnosis method of multiple transformers based on redundant information is proposed to distinguish whether the source of the fault is the transformer itself or the system disturbance.Based on this,a comprehensive detection system for electronic transformers is designed.