摘要
物联网(Internet of Things,IoT)设备故障预测是降低设备故障损失的重要环节。现有的预测算法多依赖统计分析或机器学习方法,常忽视时序数据的动态变化特性。研究提出了一种基于时序图神经网络(Temporal Graph Neural Network,TGNN)的故障预测模型。该模型结合图神经网络(Graph Neural Network,GNN)与时间序列分析技术,能够同时捕捉设备间的时空依赖关系和动态交互特征。通过时序图建模和图卷积操作,模型可准确刻画设备状态的时序变化,为故障预测提供更可靠的依据。
Internet of Things(IoT)equipment fault prediction is a critical component for reducing equipment fault losses.Current predictive algorithms predominantly rely on statistical analysis or machine learning methods,often overlooking the dynamic variation characteristics of time-series data.The study proposes a fault prediction model based on temporal graph neural networks(TGNN),which integrates graph neural network(GNN)technology with time series analysis to simultaneously capture spatiotemporal dependencies and dynamic interaction characteristics between devices.By employing temporal graph modeling and graph convolution operations,the model can accurately depict the temporal evolution of equipment status,thereby providing more reliable bases for fault prediction.
作者
张博
ZHANG Bo(Heilongjiang University of Business and Technology,Harbin Heilongjiang 150025,China)
出处
《信息与电脑》
2025年第19期33-35,共3页
Information & Computer
关键词
时序图神经网络
物联网
故障预测
图神经网络
时间序列分析
temporal graph neural network
Internet of Things
fault prediction
graph neural network
time series analysis
