Short-time traffic flow prediction method based on machine learning

The invention discloses a short-time traffic flow prediction method based on machine learning, the method comprising the steps of: S1, acquiring ground sensing coil data; S2, dividing the ground sensing coil data into different types of sections according to a network topological relation of coil groups, and based on a Daganzo model, calibrating traffic flow parameters for the different types of sections; S3, using a spatial-temporal association relationship as an eigenvector; and S4, performing speed value short-time prediction on the different types of sections by using the eigenvector as the input characteristic of an LSTM model. The short-time traffic flow prediction method based on machine learning classifies the sections, performs speed prediction by using the LSTM model according to the historical traffic flow characteristics of upstream and downstream nodes of a prediction node, and ensures the prediction accuracy and the prediction speed. 本发明公开了一种基于机器学习的交通流短时预测方法，所述方法包括：S1、采集地感线圈数据；S2、将地感线圈数据按照线圈组的网络