Research On Fault Detection System And Algorithm Of Oil-immersed Power Transformer

Transformer plays an important role in power transmission and voltage switching.Real time perception of transformer operation status is of great significance to eliminate potential faults and power grid security management.Now,our country in the processing of sensing and signal technology level has been greatly improved,the application of this technology to the transformer fault detection system can achieve very effective online detection effect,so effectively improve the accuracy of transformer fault judgment is the focus of the current research.This article first makes a brief summary of the common types of transformer faults,analyzes the composition and generation mechanism of the dissolved gas in the main faults(electrical and thermal faults)of transformer oil,and selects an appropriate gas-sensitive sensor to collect the characteristic gas composition in insulation oil in real time to establish a chromatographic data set.Then,an intelligent diagnosis method based on Dissolved Gas Analysis Technology(DGA)is proposed to perform real-time online fault diagnosis of transformers.From the introduction of multiple followers leadership model,adaptive adjustment strategy and food Gaussian mutation strategy three aspects to improve salp group algorithm,after verify the validity of the improved performance to optimize the BP neural network weights and threshold of transformer fault diagnosis simulation model is established,using the state grid corporation of failure data for simulation analysis.Finally,through comparative analysis of Particle Swarm Optimization algorithm BP neural network(PSO-BP),Moth Flame Optimization algorithm BP neural network(MFO-BP),Salp Swarm Algorithm Optimization BP neural network(SSA-BP)and Improved Salp Swarm Algorithm Optimization BP neural network(ISSA-BP)in the transformer fault detection.The comparison results show that the fault diagnosis system established in this paper has a diagnostic accuracy rate of more than 90%,and the accuracy and speed of fault diagnosis are improved to different degrees compared with other models,which verifies the effectiveness of the intelligent diagnosis model proposed in this article.