Related Experiment Video
Updated: Jun 13, 2025
14:55
Evaluating the Effect of Roadside Parking on a Dual-Direction Urban Street
Published on: January 20, 2023
3.2K
Road traffic noise prediction model based on artificial neural networks
Óscar Acosta1,2, Carlos Montenegro1, Rubén González Crespo2
1Universidad Distrital Francisco José de Caldas, Carrera 7 40b 53, Bogotá, 111711, Cundinamarca, Colombia.
Heliyon
|September 12, 2024
Summary
This study introduces a machine learning model to predict road traffic noise in Bogota, Colombia. The Multilayer Perceptron regressor demonstrated superior accuracy compared to traditional statistical models.
Area of Science:
- Environmental Science
- Urban Planning
- Data Science
Background:
- Road traffic noise is a significant urban environmental concern.
- Accurate prediction models are crucial for noise mitigation strategies.
Purpose of the Study:
- To develop and evaluate a machine learning model for predicting road traffic noise in Bogota, Colombia.
- To compare the performance of a Multilayer Perceptron (MLP) regressor against traditional statistical models.
Main Methods:
- Collected audio and video data through field measurement campaigns.
- Utilized vehicle capacity, speed, flow type, and lane count as input variables.
- Trained and compared five machine learning models, optimizing hyperparameters via mesh search.
Main Results:
- The Multilayer Perceptron (MLP) regressor achieved the best performance.
- The MLP regressor demonstrated a Mean Absolute Error (MAE) of 0.86 dBA on test data.
- MLP regressor outperformed classical statistical models in error and fit indicators.
Conclusions:
- The proposed MLP regressor is an effective tool for road traffic noise prediction.
- Machine learning models offer improved accuracy over traditional methods for urban noise assessment.
