Ensemble Hardness Harmonize Classifier for Imbalanced Fault Classification

Fault diagnosis is a crucial step to ensure the normal operation of machines and equipment. However, realworld fault data usually presents an imbalanced distribution and low quality, which makes the classification task difficult. To address this problem, this paper delves into the nature of imbalanced classification and finds that in addition to class imbalance, other difficulties such as noise and class overlap also affect the learning of effective classifiers. Therefore, this paper proposes a new imbalanced classification framework that aims to generate powerful integrated classifiers by downsampling and adaptively adjusting the data difficulty. In this paper, extensive experiments are conducted using fault data from oil pumping machines and data from wind blade machine sheets. The results show that the new framework proposed in this paper not only excels in computational efficiency, but also provides robust performance even with highly overlapping fault classes and extremely skewed data distributions. It is important to note that the method in this paper can be easily adapted to existing learning methods such as C4.5, SVM, GBDT, and neural networks to improve their performance on imbalanced fault diagnosis data. This research provides an effective solution for imbalanced classification for practical applications in the field of fault diagnosis.