Image classification method for monitoring state of electric transmission and transformation equipment on line

Abstract

The invention provides an image classification method for monitoring a state of electric transmission and transformation equipment on line. The method comprises the following steps: labeling positive and negative samples; extracting a characteristic data value, and forming an original training sample set; obtaining a classification model; performing cross validation on the classification model; classifying state images of electric transmission and transformation equipment. According to the image classification method, effective classification models can be trained aiming at cost-sensitive and class-unbalanced characteristics of state monitoring images of the electric transmission and transformation equipment, the operating state of the electric transmission and transformation equipment is effectively and automatically judged in real time through identification and classification of the state image of the equipment, abnormal equipment is rapidly and accurately found, and the effectiveness and intelligent level of inspection information processing of the electric transmission and transformation equipment are improved.

Description

technical field [0001] The invention relates to a classification method, in particular to an image classification method used for on-line monitoring of the state of power transmission and transformation equipment. Background technique [0002] With the rapid development of electric power construction and the continuous expansion of power grid scale, the overhaul and maintenance of power grid equipment are becoming more and more frequent. Power transmission and transformation equipment plays an important role in power operation. The condition monitoring and fault diagnosis of power transmission and transformation equipment are the basis for ensuring the safe and stable operation of the power grid. [0003] Equipment status monitoring includes online monitoring, offline detection and testing when necessary, and all means of obtaining operating status data that are not in direct contact with operating equipment (such as aerial photography and infrared monitoring, etc.). Accord...

AI Technical Summary

Problems solved by technology

Traditional machine learning image classification methods include support vector machines (SVM), decision trees, multi-layer neural networks, etc., but these methods do not take into account the cost-sensitive and class imbalance problems when solving classification problems

There is a problem of category imbalance in the monitoring images of the state of power transmission and transformation equipment, and the images of the normal category are far more than the images of the abnormal category; at the same time, there is also a cost-sensitive problem in the judgment of the status of the equipment, and the normal category images are judged as abnormal It is different from the cost of judging abnormal images as normal

Images