Passenger flow prediction method based on self-attention personalized enhanced graph convolutional network

Abstract

The invention discloses a passenger flow prediction method based on a self-attention personalized enhanced graph convolutional network, and the method comprises the steps: collecting the historical passenger flow data of a subway network, and constructing a subway network adjacency matrix at different moments; the subway network adjacency matrix and the historical passenger flow volume data of each subway station are input into a personalized enhanced graph convolutional neural network P-GCN, spatial features in a passenger flow data set are extracted, and a trainable diagonal matrix is defined in the graph convolutional neural network P-GCN; constructing position codes, inputting the position codes and spatial features in the passenger flow data set into the multi-head self-attention network, and calculating a query matrix, a key matrix and a value matrix; calculating new passenger flow data of each subway station by using the query matrix, the key matrix and the value matrix V; and performing standardization processing on the new passenger flow data, optimizing the self-attention personalized enhanced graph convolutional network, and outputting a passenger flow prediction result. The passenger flow prediction accuracy is obviously improved, and the passenger flow in a period of time in the future is effectively predicted.

Description

technical field [0001] The invention relates to the fields of deep learning and intelligent transportation, in particular to a passenger flow prediction method based on a self-attention-based personalized enhanced graph convolutional network. Background technique [0002] Nowadays, artificial intelligence technology is developing rapidly. At the same time, with the massive application of cheap traffic sensor technology and the explosive growth of traffic data, human beings have entered the era of traffic big data and intelligent transportation. The intelligent transportation system aims to establish a complete set of traffic information service and traffic management control system, which can alleviate the traffic congestion problem. For example, the intelligent transportation system can test the station pressure for the passenger flow prediction of the rail transit station, and it is reasonable according to the predicted passenger flow data. Plan the line timetable to effec...

AI Technical Summary

Problems solved by technology

However, the original GCN has the problem that the predicted value is smooth, that is, the peak value of the predicted value is not obvious

Images