摘要
机电系统故障诊断与预测面临早期故障识别难、多源信息融合复杂等挑战。提出一种基于深度学习的智能诊断框架,采用多模态特征提取与融合网络对振动、电流、温度等异构传感器数据进行集成分析,构建端到端模型,提升故障分类的准确性与泛化能力。同时,基于Transformer架构设计退化趋势预测方法,有效捕捉系统长期性能演变规律。案例验证表明,该方法在故障诊断准确率和剩余使用寿命(Remaining Useful Life,RUL)预测精度上均优于传统方法,为实现预测性维护提供有效工具。
The diagnosis and prediction of faults in electromechanical systems face challenges such as difficulty in early fault identification and complex fusion of multi-source information.A deep learning based intelligent diagnosis framework is proposed,which integrates and analyzes heterogeneous sensor data such as vibration,current,and temperature using multimodal feature extraction and fusion networks,and an end-to-end model is constructed to improve the accuracy and generalization ability of fault classification.Meanwhile,a degradation trend prediction method based on Transformer architecture is designed to effectively capture the long-term performance evolution of the system.Case studies have shown that this method outperforms traditional methods in both fault diagnosis accuracy and Remaining Useful Life(PUL)prediction accuracy,thus providing an effective tool for achieving predictive maintenance.
作者
王栋
李伟
霍鑫睿
黄炎彬
WANG Dong;LI Wei;HUO Xinrui;HUANG Yanbin(Shanxi General Aviation Polytechnic,Datong 037000)
出处
《现代制造技术与装备》
2026年第1期117-119,共3页
Modern Manufacturing Technology and Equipment
关键词
智能故障诊断
深度学习
预测性维护
intelligent fault diagnosis
deep learning
predictive maintenance
