A safety production early warning system based on multi-source heterogeneous data federated learning

Abstract

The present invention provides a safety production early warning system based on federated learning of multi-source heterogeneous data, including: a central decision server and multiple node clients, a central model is established on the central decision server, and each central model processes a kind of data; Build a local model on the node client, use local data for model training, and generate local model parameters; select the corresponding central model, and upload the local model parameters to the corresponding central model; each central model receives multiple local model parameters, Establish a global parameter model; the node client downloads the global parameter model until the central model converges or reaches the predetermined training round; the central model after training is delivered to each node client; the node client generates recognition results and implements early warning. The present invention solves the problem of independent administration of safe production in each park, realizes joint prevention and control of safe production, jointly solves problems related to each other, and ensures data privacy of park enterprises.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a safety production early warning system based on federated learning of multi-source heterogeneous data. Background technique [0002] The current safety production situation and goal of the chemical park is to cooperate with all enterprises in the park to monitor and warn of safety production. But the reality is that the safety production data of each enterprise in the park is owned by each enterprise, and data sharing involves data privacy and security issues, which does not conform to the current data development trend. Although traditional federated learning algorithms can solve the problem of data value sharing, they are helpless in the face of the multi-dimensional and multi-type characteristics of the professions and data involved in safety production prevention and control. Contents of the invention [0003] The technical problem to be solved by the pre...

AI Technical Summary

Problems solved by technology

Although the traditional federated learning algorithm can solve the problem of data value sharing, the traditional federated learning algorithm is helpless in the face of the multi-dimensional and multi-type characteristics of the professions and data involved in safety production prevention and control

Images