fault diagnosis of power grid equipment based on artificial intelligence with limited data

Solving the problem of diagnosing old power grid equipment is of high scientific importance and relevance in the modern energy sector. although old equipment may be reliable and continue to function, it is still subject to wear and tear and possible malfunctions, which can lead to accidents, power failures and even potential environmental hazards. the use of artificial intelligence technologies has led to major changes, but less models work well when implemented into a technological process, since they require large amounts of training data that the industry is not ready to provide. the article is devoted to solving the problem of machine learning model learning on a very limited amount of data. the description of the method combining two ideas is given: the use of a neural network of comparison, and an ontological knowledge base, the content of which is enriched by the r-gcns model. the formation of an ontological knowledge base about defective states of electrical equipment plays a key role in the development of new scientific topics. it allows you to systematize and disseminate knowledge about defects in electrical equipment in general, and defective insulation conditions of a power transformer in particular, contributing to the development of innovations and improving the quality of the planning process for maintenance and repair of electrical equipment. to accumulate data on the state of isolation of a power transformer in the object space of a relational graph, and restore missing facts, i.e. the subject-predicate-object triplet, r-genes belonging to the recently appeared class of neural networks working with graphs will be used. to solve the problem of insufficient data, a neural network matching algorithm is proposed that uses the similarity function to find the relationship between two compared instances and classify new data classes without retraining the network. in general, the proposed solution is aimed at automating the prediction of the state of the power transformer and optimizing the planning of maintenance work, which ultimately should reduce downtime and increase the reliability of power grids.