Transformer fault diagnosis method based on deep forest model

Abstract

The invention discloses a transformer fault diagnosis method based on a deep forest model. The method comprises the steps that a non-coding ratio of analysis data of dissolved gas in transformer oil is taken as a characteristic parameter of the deep forest model, and sample data is divided into a training set and a test set; and then a deep forest model DF is constructed, the deep forest model DFextracts more feature information from multi-dimensional data of a transformer fault through multi-granularity scanning, and the effect of diagnosing and identifying the fault type of the transformeris optimal through cascade forest training. According to the method, the fault diagnosis accuracy of the transformer is effectively improved, and a reliable basis is provided for operation and maintenance personnel to correctly judge the operation condition of the transformer.

Description

Technical field [0001] The invention relates to the technical field of power equipment monitoring, in particular to a transformer fault diagnosis method based on a deep forest model. Background technique [0002] With the development of energy interconnection, transformer failure will endanger the safe and stable operation of the entire power system. Therefore, a quick and accurate understanding of the fault type of the transformer, so as to perform maintenance work, can provide an important guarantee for the normal operation of the power system. [0003] Dissolved Gas Analysis (DGA) is mainly used in the online monitoring of oil-immersed transformers. Based on DGA characteristic gas for data correlation analysis, domestic and foreign researchers have proposed three-ratio method, Rogers ratio method, Dornenburg ratio method, and electrical research method, but the traditional DGA method only gives the threshold of fault diagnosis. , Cannot faithfully represent the law between cha...

AI Technical Summary

Problems solved by technology

For example, the expert system cannot learn independently, the work efficiency is low, and it is difficult to obtain accurate diagnosis results; the neural network has strong self-learning ability, but requires a large amount of sample data for training, hyperparameter adjustment is complicated and the learning cycle is long, and it is easy to fall into a local minimum ;The decision tree relies on optimizing the local optimum to achieve the overall optimum, and it is difficult to ensure that it returns to the global optimum, and it is easy to overfit. The above diagnostic methods all have the problem that it is difficult to effectively deal with high-dimensional data and feature information extraction.

Images