Interval Short-Term Traffic Flow Prediction Method Based on CEEMDAN-SE Nosie Reduction and LSTM Optimized by GWO

Table 1

Recent works on intelligent transportation system.

Number	Solved problem	Model name
1	Sensor data is analyzed using crow search algorithm optimized long-short-term memory to correctly identify drivers [2].	Crow search algorithm optimized long-short-term memory (CSA-LSTM)
2	Spatio-temporal individual mobility graph encoding network with group mobility assistance (SIGMA) is proposed to encode individual mobility behavior, which enables recommendation of new locations [3].	Spatio-temporal individual mobility graph encoding network with group mobility assistance (SIGMA)
3	This work proposes a deep learning-based traffic safety solution in 5G intelligent transportation systems that can effectively predict drivers’ intention to change lanes [4].	(1) Lane-change intention recognition based on an LSTM and historical driving-track data; (2) Lane-change intention recognition based on an LSTM and natural-driving data; (3) Recognition of lane-change intention based on decision layer fusion
4	This work presents an edge node deep learning-based traffic flow detection scheme that combines vehicle detection and vehicle tracking algorithms and is deployed to the edge device Jetson TX2 platform [5].	A vehicle detection network based on improved YOLOv3 and a vehicle tracking network based on the improved DeepSORT
5	This work proposes a dynamic and intelligent traffic light control system (DITLCS) that dynamically adjusts traffic light durations by analyzing real-time traffic information, which improves the efficiency of traffic light control systems [6].	Deep reinforcement learning and fuzzy inference system
6	This work presents a radial basis function neural network algorithm based on quantum particle swarm optimization (QPSO) strategy for traffic flow prediction in intelligent transportation system (ITS) [7].	Quantum particle swarm optimization (QPSO) strategy