Abstract:Urban traffic control has become a big issue to help traffic management in recent years. With data explosion, Intelligent Transportation System (ITS) is developing rapidly. ITS is an advanced data-based method for traffic control, which requires timely and effective information supply. This research aims at providing real-time and accurate traffic flow data by intelligent prediction method. Applying multiple road traffic flow data of the Caltrans Performance Measurement System (PeMS) and separating the time series, the mechanism of spatial-temporal differences was taken into consideration. Based on the basic Long Short-Term Memory (LSTM) model, an improved LSTM model with Dropout and Bi-structure (Bi-LSTM) for traffic flow prediction was presented. In the prediction process, we applied three models including the improved Bi-LSTM model, Gated Recurrent Unit (GRU) model and Linear Regression in the experiment, and made a comparison from aspects of model structure complexity, operating efficiency and prediction accuracy. To validate the portability of the prediction model, the features of traffic flow from different datasets were further analyzed. The experimental results show that the improved Bi-LSTM model performs best in traffic flow prediction with comprehensive rationality, which reaches an accuracy of about 92% when considering temporal differences. Particularly, the specific factors of traffic situations and locations which is more applicable to be predicted by the improved Bi-LSTM model are summarized considering spatial differences. This research proposes an advanced and accurate model to provide real-time and short-term traffic flow prediction data, which is of great help to intelligent traffic control. Considering the mechanism between model and road traffic properties, the results suggest that it is more applicable in urban commercial area.

A Time Series Prediction Model Based on Long Short-Term Memory Networks

Urban Traffic Flow Forecasting Based on Memory Time-Series Network.

Hourly and Daily Urban Water Demand Predictions Using a Long Short-Term Memory Based Model

A Novel Model for the Prediction of Long-Term Building Energy Demand: LSTM with Attention Layer

Short-Term Traffic Flow Prediction for Urban Road Sections Based on Time Series Analysis and LSTM_BILSTM Method

Predicting Short-Term Traffic Flow by Long Short-Term Memory Recurrent Neural Network.

Short-term Travel Time Prediction by Deep Learning: A Comparison of Different LSTM-DNN Models

Short-term Passenger Flow Prediction of Subway Based on Neural Network LSTM Algorithm

An LSTM-Based Method Considering History and Real-time Data for Passenger Flow Prediction

Short-Time Traffic Flow Prediction Based on K-means++ and LSTM.

Deep Learning with Bidirectional Long Short-Term Memory for Traffic Flow Prediction

Short-Term Prediction of Urban Rail Transit Passenger Flow in External Passenger Transport Hub Based on LSTM-LGB-DRS

A Spatial-Temporal Approach for Traffic Status Analysis and Prediction Based on Bi-LSTM Structure

Comparison of ARIMA and LSTM in Predicting Structural Deformation of Tunnels during Operation Period

Short-Term Prediction of Bus Passenger Flow Based on a Hybrid Optimized LSTM Network

Traffic flow prediction using bi-directional gated recurrent unit method

A Combination Prediction Model for Short Term Travel Demand of Urban Taxi

Short-term Prediction of Bike-sharing Usage Considering Public Transport: A LSTM Approach

A Network Traffic Prediction Method Based on LSTM

A Machine Learning Based Method for Short-term Urban Rail Transit Passenger Flow Prediction

Passenger flow prediction and management method of urban public transport based on SDAE model and improved Bi-LSTM neural network