摘要详情

ID / 提交时间

203 / 2025-06-14 09:41:08

标题

Research on Motor Fault Diagnosis Based on Improved Convolutional Neural Network of Random Forest under Small Sample Conditions

关键字

small sample,CNN,random forest,fault diagnosis

主题及专题

Special Sessions > SSL 24 Annual TEPEN & Silk Road Collaboration Symposium

状态

终稿

作者

Wenjie Liu / National University of Defense Technology

Jiajie Hao / Hunan University

Yi Yang / National University of Defense Technology

Guoji Shen / National University of Defense Technology

摘要

In response to the limitations of traditional convolutional neural networks in motor condition monitoring under small sample conditions, this study proposes an improved convolutional neural network model based on random forests（CNN-RF）, which utilizes Random Forests to replace the softmax classification layer of the original CNN, thereby integrating Random Forest methods with Convolutional Neural Networks. Utilizing the bearing dataset from Huazhong University of Science and Technology, the research generated image sets under small-sample conditions using continuous wavelet transform. The proposed model demonstrates robust classification performance in data-scarce scenarios, achieving 97.38% accuracy with only 10% of the data.