Related Experiment Video
Updated: Aug 14, 2026
14:55
Evaluating the Effect of Roadside Parking on a Dual-Direction Urban Street
Published on: January 20, 2023
Enhancing Road Sensor Data Fidelity via FMD Decomposition for Accurate Traffic Flow Prediction Using EBWO-Optimized
Jixiao Jiang1, Anastasia Feofilova1, Ivan Topilin1
1Department of Transportation and Traffic Management, Don State Technical University, 344002 Rostov-on-Don, Russia.
Sensors (Basel, Switzerland)
|August 13, 2026
Summary
This study introduces a novel hybrid framework for accurate traffic flow prediction, significantly reducing errors caused by noisy sensor data in intelligent transportation systems (ITSs). The method enhances prediction accuracy by over 30% compared to existing models.
Area of Science:
- Intelligent Transportation Systems (ITSs)
- Data Science
- Signal Processing
Background:
- Resource-constrained ITSs face challenges with noisy road sensor data, impacting traffic flow prediction accuracy.
- Non-stationary noise and outliers degrade the reliability of traffic data.
Purpose of the Study:
- To develop a robust hybrid prediction framework for accurate traffic flow prediction.
- To address data quality issues caused by noise and outliers in ITS data.
Main Methods:
- Frequency Mode Decomposition (FMD) for adaptive noise separation.
- Enhanced Beluga Whale Optimization (EBWO) for spatiotemporal characteristic exploration.
- Logistic-Gaussian Circle-based Bidirectional Gated Recurrent Unit (LGC-BiGRU) for enhanced periodic pattern detection.
Main Results:
- The proposed framework significantly reduces Mean Absolute Error (MAE) by 30.6% and Root Mean Square Error (RMSE) by 29.9%.
- Demonstrated superior performance compared to state-of-the-art models in traffic flow prediction.
Conclusions:
- The hybrid framework effectively handles complex noise in traffic data.
- The approach is effective and feasible for real-world traffic flow prediction in ITSs.