Transfer learning fault diagnosis method considering multi-element factor situation evolution for distribution transformer

Abstract

The invention provides a transfer learning fault diagnosis model of a distribution transformer. The transfer learning fault diagnosis model is implemented according to the steps of (1) dividing statequantity affecting a running state of the distribution transformer into dynamic state quantity, quasi-dynamic state quantity and static state quantity, and building a running state evaluation index system of the distribution transformer; (2) performing binary quantization on the index state quantity, digging an incidence relation with a fault by an Apriori algorithm, and extracting key index statequantity of inducing a fault of the transformer; (3) introducing a Tanimoto coefficient, and transferring effective auxiliary fault data to a target distribution transformer; and (4) performing iterative solution on weights of target fault data and auxiliary fault data by a transformer learning algorithm TrAdaBoost to obtain a fault diagnosis model of the distribution transformer for fault diagnosis of the target distribution transformer. Auxiliary fault information of the distribution transformer is transferred to the target distribution transformer, and the problem brought to fault diagnosis of the distribution transformer due to a few pieces of single fault data is solved very well.

Description

Technical field [0001] The present invention belongs to the field of fault diagnosis of distribution transformers, and more specifically, is a method for fault diagnosis of distribution transformer migration learning that considers the evolution of multiple factors. Background technique [0002] In the distribution network, there are a huge number of distribution transformers (distribution transformers), and ensuring their safety is the basis for the stable and reliable operation of the power grid. Accurate perception of their status, accurate diagnosis of faults, and timely investigation of risks are essential for ensuring the reliability of power supply, It is of great significance to realize risk early warning and reduce the probability of accidents. [0003] In recent years, with the rapid development of big data, data mining and other technologies, it has been widely used in transformer fault diagnosis. The research mainly focuses on the intelligent extraction of fault featur...

AI Technical Summary

Problems solved by technology

[0004] In view of the above defects or improvement needs of the prior art, the present invention aims to provide a distribution transformer migration learning fault diagnosis method that considers the situation evolution of multiple factors, and solves the problem of distribution transformer fault diagnosis for the problem of few single fault data of distribution transformers

Images