Today the world is facing the exploration of artificial intelligence, which is a vast area imitating the faculties that humans learn, gradually improving its accuracy. Machine learning can be adopted in image recognition application which is a technique used for cataloging objects or features in the environment of digital image. However, due to the concern of data privacy, devices are not willing to transmit their raw data to base stations for machine learning training. In the paper, we adopt federated learning for mobile devices training and data privacy. By using permissioned Hyperledger fabric blockchain, we propose a model of federated learning based Industrial Internet of Things (IIoT) to protect Industrial information by sharing data through a privacy preserving module of differential privacy. Simulation results show that the approach maintains 96.15% accuracy in the shared data for ε = 0.5 which provides a sufficient privacy preservation guarantee.