Research On Information Fusion Of Multi-source Monitoring Data Of Power Transformers

In the era of rapid development of the power industry,power transformers are becoming more and more important to maintain the safety and stability of power systems.Once the transformer failure has unpredictable consequences for manpower and material resources,it is imperative to maintain the safe and stable operation of the transformer.How to judge the potential faults of transformers in a timely,early and accurate manner has always been one of the topics that researchers have been studying for a long time.Because the data used in the research of traditional power transformer fault diagnosis has a single source of data,and the transformer data There are structural,semi-structured and unstructured phenomena,which make it difficult to process.As a result,there are still some shortcomings in the accuracy effect,which can not meet the needs of people well.Therefore,this paper proposes against this background.A method of multi-source information fusion technology in the field of transformer fault diagnosis.This paper first summarizes the research status of transformer fault and multi-source information fusion,and makes a corresponding research on the knowledge of multi-source information fusion.The application scenarios of information fusion technology in military and civil affairs are discussed.The model,construction and training methods of deep belief neural network algorithm are introduced,and the simulation training is carried out.The simulation results show that the accuracy of the model changes with the number of iterations.Then the paper constructs a model and algorithm flow chart of transformer fault diagnosis based on deep belief neural network and D-S evidence theory algorithm.Taking the dissolved gas and partial discharge in the transformer oil as the input data of the model,by combining the deep belief network with the D-S evidence theory,and through repeated training network and tuning,the algorithm accuracy is best.Finally,the experimental results are compared with the traditional neural network and D-S algorithm,which proves the rationality of the proposed algorithm.The experiment shows that the multi-source information fusion technology proposed in this paper is applied to the correctness of transformer fault diagnosis,and solves the single characteristics of the traditional method,which provides a decision basis for ensuring the safe operation of the power system.