Key node recognition of intelligent transportation network based on improved deep belief network

Aiming at the problems of low false recognition rate, missed recognition rate, and long recognition time in current recognition methods, a key node recognition method of intelligent transportation network based on improved deep belief network is proposed. Firstly, construct a BA scale-free network model to characterize the power-law distribution characteristics of intelligent transportation networks. Secondly, based on the modeling results, a Meta-ST-I model was designed, which integrates spatiotemporal features through static embedding and dynamic embedding, and effectively fills in missing information using spatial attention mechanism and temporal attention mechanism. Finally, an extreme learning machine (ELM) is used to improve the deep belief network, and the filled data is input into the improved deep belief network to achieve key node recognition. Experimental results show that the average false positive rate of the proposed method is 4.63%, the average false negative rate is 5.39%, and the recognition time is stable between 0.25s~0.68s.