Acoustic Emission Based Fault Detection of Substation Power Transformer

Xin Ma,Jian Shi,Hailiang Xiong,Yu Luo

doi:10.3390/app12052759

Abstract

Fault detection of Substation Power Transformer by Non-contact measurement is important for the safety of machines, instruments, and human beings. To make non-contact measurement as convenient as possible, it is desirable that efficient algorithms based on AE (acoustic emission) discrimination are developed. This paper presents a system for quick and effective fault detection of substation power transformer, based on AE signals collected by non-contact single microphones. In the experiment, collected data were preprocessed in multiple ways and three machine learning algorithms were designed based on classifiers (Convolutional Neural Network (CNN), support vector machine (SVM), and k-nearest neighbors (KNN) algorithm) trained and tested by a tenfold cross-validation technique. After comparison among the designed classifiers, the results show the two-dimensional principal component analysis (2DPCA) preprocess combined with SVM achieved the best comprehensive effectiveness.

Highlights

The effective signals are transferred into a time-frequency representation to yield [23]: Xn ejω x(m)ω(n where x(n) is the signal received after endpoint detection and ω(n) is the window function, which shifts the AE signals signal by one step length on the time axis
In PCA analysis (Figure 4a), the first two coefficients account for a larger proportion, while In 2DPCA analysis (Figure 4b), the first coefficient account for a larger proportion, With the μ increase, the number of eigenvectors 2DPCA analysis change more slowly
The PCA projection matrix is the size of 171 × 2, the feature vector dimension of each frame is reduced to 2; 2DPCA projection matrix is 9 × 1 vector, the feature matrix is reduced from 39 × 9 to a 39 × 1 vector

Summary

Introduction

Non-contact measurement [1,2], nondestructive testing [3–7] are the key components of robot automatic inspection technology, especially faced with some dangerous, precious, or tender tested objects. Based on some special sensors designed according to one or several physics theories, some features related to the inspected devices were extracted from the recorded signals, using these features the computer system undertakes the analysis and diagnose work. Many systems in which the non-contact measurement or nondestructive measurement were applied have already been deployed to the market and have played useful roles in everyday life. Many important results of non-contact measurement have already been reported and many systems have already been applied in the field, the technologies and theories are still far from being mature. A successful application of AE signals in the field of non-contact measurement can improve security and decrease the occupation of resources [8,11,12]

Objectives

Methods

Results

Discussion

Conclusion