深度学习在故障诊断领域中的研究现状与挑战

发布时间：2018-03-17 15:57

本文选题：深度学习　切入点：复杂工业系统　出处：《控制与决策》2017年08期 　论文类型：期刊论文

【摘要】：现代工业系统已呈现出向大型化、复杂化的方向发展,使得针对工业系统的故障诊断方法遇到一系列的技术难题.近年来,深度学习(deep learning)在特征提取与模式识别方面显示出独特的优势与潜力,将深度学习应用于解决复杂工业系统故障诊断的研究已初现端倪.为此,首先介绍几种典型的基于深度学习方法实现工业系统故障诊断方法;然后对基于深度学习实现故障诊断的主要思想和建模方法进行描述;最后总结和讨论了复杂工业系统故障的特点,并探讨了深度学习在实现复杂工业系统故障诊断方面所面临的挑战,展望了未来值得继续研究的方向.
[Abstract]:The modern industrial system has developed towards the direction of large scale and complexity, which makes the fault diagnosis method for industrial system meet a series of technical difficulties. Deep learning has shown unique advantages and potential in feature extraction and pattern recognition. The application of deep learning to fault diagnosis of complex industrial systems has begun to emerge. This paper first introduces several typical deep learning methods to realize fault diagnosis of industrial system, then describes the main ideas and modeling methods of fault diagnosis based on depth learning. Finally, the characteristics of complex industrial system faults are summarized and discussed, and the challenges of deep learning in fault diagnosis of complex industrial systems are discussed, and the future research directions are prospected.
【作者单位】：重庆大学自动化学院;电力传输设备与系统安全国家重点实验室;航天发射场可靠性技术重点实验室;
【基金】：国家自然科学基金项目(61275162) 重庆市基础与前沿研究计划项目(cstc2016jcyjA0504)
【分类号】：TP18;TP277

【参考文献】