CFDM-IME: A Collaborative Fault Diagnosis Method for Intelligent Manufacturing Equipment

The stability of the intelligent manufacturing industry will directly affect the development of the social economy. The privacy of data among different smart factories (SF) leads to a lack of generalization of existing deep learning-based fault diagnosis methods. In order to solve the problems existing in the fault diagnosis method, this paper combines blockchain, federated learning, and deep learning to propose a collaborative fault diagnosis method for intelligent manufacturing equipment (CFDM-IME). Specifically, firstly, a fault diagnosis model based on LSTM is proposed to realize local model training of local intelligent manufacturing equipment. Then, a domain parameter aggregation method based on a federated average is proposed to realize the aggregation of internal model parameters of smart factories. Then, the parameter collaborative optimization smart contract is designed and implemented to achieve the aggregation of global parameters. Finally, we conduct simulation experiments on the proposed method. Theoretical and simulation experiments prove that our proposed architecture is feasible.