This paper proposes fault recovery mechanism for industrial internet of things (IIoTs) using fog computing platform. The platform consists of three layers: end devices layer, fog layer, and cloud layer that connected by wireless network. Fog layer is added to minimize latency of the system. Then, machine learning algorithm is used to make decisions fast and improve the performance of the system. The proposed system mechanism is separated into four processes: stream data, fault detection, fault recovery, and dispatching result. Finally, the proposed system model using fog computing platform is described clearly. Moreover, the advantages of mechanism compared with related research.