Transformer online monitoring and fault diagnosis method

Abstract

The invention discloses a method applied to online monitoring and fault diagnosis of a transformer. The method comprises four steps. The method comprises the following steps of: 1, acquiring instantaneous values of n paths of signals such as vibration, current, voltage, temperature and the like of different measuring points or different phases of a transformer by using various sensors; 2, respectively processing the acquired n paths of signals by discrete S transformation, and forming a complex matrix of n channels by calculation results; 3, carrying out modulo operation on each element in then-channel complex matrix to obtain an n-channel real matrix, namely a feature image of the signal; and 4, inputting the feature image of the signal into the trained unsupervised deep learning framework to obtain a diagnosis result.

Description

technical field [0001] The invention relates to a fault diagnosis technology of a transformer, in particular to an online monitoring and fault diagnosis method of a transformer, and belongs to the technical field of power electronics. Background technique [0002] The power transformer is one of the most important equipment in the power grid, and maintaining the normal operation of the transformer is related to the safety of the entire power grid. In operation, however, transformers are subjected to electrical, chemical, mechanical, thermal and environmental stresses that can cause winding and core failures. Transformer faults have a great impact on the power grid. Once a sudden accident occurs during operation of a large power transformer, it may cause a large-scale power outage, causing serious adverse social impacts and economic losses. Moreover, the maintenance cost of power transformers is high and the cycle is long, and various economic losses are also very huge. [...

AI Technical Summary

Problems solved by technology

Traditional deep learning methods require a large amount of normal and abnormal data for training, but transformers have many types and are expensive, and the probability of failure in actual operation is low, so normal data is redundant and abnormal data is scarce

Images