Abstract: To address the issues of node overload and failure caused by resource constraints in industrial Internet of Things (IIoT) edge computing environments, this paper proposes a fault-tolerant method based on edge intelligence, termed SGAT-GAN (Sparse Graph Attention-based Generative Adversarial Network). The proposed method integrates a fault prediction mechanism and employs a Fault Embedding Encoder (FEE) to achieve real-time fault detection, classification, and prediction, thereby enhancing the system’s fault-tolerance capability. Furthermore, by combining a Generative Adversarial Network (GAN) with a preemptive migration strategy for optimal fault recovery, the method provides an efficient and intelligent solution for edge computing systems in the IIoT. Experimental results conducted in a Raspberry Pi–based edge environment demonstrate that the proposed method outperforms four state-of-the-art approaches—Proactive Coordinated Fault Tolerance (PCFT), Clustering-based Multiple Objective Dynamic Load Balancing technique (CMODLB), Self-supervised Deep Proxy-based Fault Tolerance (DeepFT), and Preemptive Migration Prediction Network (PreGAN)—in terms of average fault detection precision and average classification accuracy. Specifically, compared with the best-performing PreGAN method, SGAT-GAN achieves improvements of 1.96 and 1.62 percentage points in these two metrics, respectively, validating its effectiveness and superiority in enhancing the reliability of IIoT edge computing systems.

Key words: Industrial Internet of Things (IIoT), edge computing, fault-tolerant method, fault prediction, Generative Adversarial Network (GAN)

摘要： 针对工业物联网(IIoT)边缘计算环境中因资源受限导致节点过载甚至故障的问题，提出一种基于边缘智能的容错方法SGAT-GAN(Sparse Graph ATtention-based Generative Adversarial Network)。该方法融合故障预测机制，通过故障嵌入编码器(FEE)技术实现故障的实时检测、分类与预测，提升系统的容错能力。在此基础上，结合生成对抗网络(GAN)和抢先迁移选择最优策略对故障进行智能补救，为工业物联网中的边缘计算系统提供一种高效、智能的解决方案。在基于Raspberry-Pi的边缘环境中进行的实验结果表明，该方法在故障平均检测精确率和平均分类准确率上优于主动协调容错(PCFT)、基于聚类的多目标动态负载均衡技术(CMODLB)、基于自监督深度代理模型的容错方法(DeepFT)和预迁移预测网络(PreGAN)4种先进方法。具体而言，相较于表现最佳的PreGAN方法，本文方法在这两项指标上分别提高了1.96个百分点和1.62个百分点，验证了该方法在提升工业物联网边缘计算系统可靠性方面的有效性与优越性。

关键词: 工业物联网, 边缘计算, 容错方法, 故障预测, 生成对抗网络