Related Experiment Video
Updated: Mar 16, 2026
16:14
Trajectory Data Analyses for Pedestrian Space-time Activity Study
Published on: February 25, 2013
14.3K
A Hybrid Short-Term Traffic Flow Prediction Model Based on Singular Spectrum Analysis and Kernel Extreme Learning
Qiang Shang1, Ciyun Lin1,2, Zhaosheng Yang1,2
1College of Transportation, Jilin University, Changchun 130022, China.
Plos One
|August 24, 2016
Summary
This study introduces a hybrid model combining Singular Spectrum Analysis (SSA) and Kernel Extreme Learning Machine (KELM) for improved short-term traffic flow prediction in intelligent transport systems (ITS). The SSA-KELM model demonstrates superior accuracy and robustness compared to existing methods.
Area of Science:
- Intelligent Transport Systems (ITS)
- Time Series Analysis
- Machine Learning
Background:
- Short-term traffic flow prediction is crucial for ITS but challenging due to data uncertainty and nonlinearity.
- Existing models often struggle with noise and complex traffic dynamics.
Purpose of the Study:
- To enhance the accuracy and robustness of short-term traffic flow prediction.
- To propose a novel hybrid model integrating SSA and KELM.
Main Methods:
- A hybrid model (SSA-KELM) was developed, utilizing Singular Spectrum Analysis (SSA) for noise filtering.
- Kernel Extreme Learning Machine (KELM) was employed for prediction, with parameters optimized by Gravitational Search Algorithm (GSA).
- Phase space reconstruction was used to determine the optimal input for the KELM model.
Main Results:
- The SSA-KELM model significantly outperformed established models like SVM, ELM, and standalone KELM in case validation.
- The proposed model demonstrated improved accuracy and enhanced robustness in traffic flow prediction.
Conclusions:
- The SSA-KELM hybrid model offers a more effective solution for short-term traffic flow prediction.
- This approach addresses the limitations of existing methods in handling noisy and nonlinear traffic data.
