Traffic flow prediction method based on bidirectionally nested LSTM neural network

Abstract

The invention discloses a traffic flow prediction method based on a bidirectionally nested LSTM neural network. The method comprises the steps of acquiring traffic flow data of a prediction road section and K most related road sections, building a road traffic flow space-time matrix data set and performing data serialization treatment; then building a bidirectionally nested LSTM neural network prediction model, defining a prediction model loss function, and combining with the training set data to complete model training; and at last, using the data of a test set as the input of the trained model, achieving the real-time prediction of the traffic flow state of the test set and defining a model assessment standard, to perform error analysis. According to the method provided by the invention,through improving the LSTM unit time hierarchical effect and the linkage between the future, historical traffic flow states and the current state, the time feature extraction capacity of the road traffic flow data is improved, and thus the prediction accuracy of the road traffic flow is improved.

Description

technical field [0001] The invention belongs to the field of traffic prediction and relates to a traffic flow prediction method based on a bidirectional nested LSTM neural network. Background technique [0002] With the continuous advancement of urban modernization, people's living standards continue to improve, and the number of vehicles per capita continues to increase. The existing urban road traffic network construction is difficult to meet the growing demand for road traffic travel. Road congestion under road traffic construction. As an important part of the intelligent transportation system, the road traffic flow prediction can not only assist the road traffic management department to control and guide the road traffic flow, but also provide a basis for people to make more reasonable travel decisions. [0003] Existing road traffic flow forecasting methods can be divided into three categories: The first category is traditional forecasting models based on mathematical ...

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the shortcomings of the low prediction accuracy of existing traffic flow prediction methods, the object of the present invention is to provide a traffic flow prediction method based on bidirectional nested LSTM neural network

Images