Decentralization satellite fault diagnosis method and device based on federated learning and block chain

The invention discloses a decentralized satellite fault diagnosis method and device based on federated learning and a block chain, and the method comprises the steps: migrating the aggregation work of a conventional federated learning model to a satellite server based on a block chain technology, proposing an excitation and verification mechanism for a satellite client node, and achieving the fault diagnosis of a satellite through a privacy protection mechanism. The information security problem of model parameter transmission in a satellite system is solved; meanwhile, the defects of the prior art in the aspects of real data acquisition, satellite client node incentive policies and privacy protection are overcome; according to the federated learning system based on block chain technology decentralization, a completely decentralized joint expectation maximization algorithm is used for training a model, the problem that traditional federated learning excessively depends on a central server is effectively solved