Open Access Open Access  Restricted Access Subscription Access

Copyright © 2024 ISEIS. All rights reserved

A Bayesian Method for Model Selection in Environmental Noise Prediction

L. Martín-Fernández1*, D. P. Ruiz1, A. J. Torija2, J. Míguez3

  1. Department of Applied Physics, University of Granada, Avda. Fuentenueva s/n, Granada 18071, Spain
  2. Institute of Sound and Vibration Research, University of Southampton, Highfield, Southampton SO17 1BJ, UK
  3. Department of Signal Theory & Communications, Carlos III University of Madrid, Avda. de la Universidad 30, Leganés, Madrid 28911, Spain

*Corresponding author. Tel: +349-58244161 Fax: +349-58243214 Email:


Environmental noise prediction and modeling are key factors for addressing a proper planning and management of urban sound environments. In this paper we propose a maximum a posteriori (MAP) method to compare nonlinear state-space models that describe the problem of predicting environmental sound levels. The numerical implementation of this method is based on particle filtering and we use a Markov chain Monte Carlo technique to improve the resampling step. In order to demonstrate the validity of the proposed approach for this particular problem, we have conducted a set of experiments where two prediction models are quantitatively compared using real noise measurement data collected in different urban areas.

Keywords: environmental noise level prediction, MAP model selection, Monte Carlo sampling, nonlinear state-space model, particle filtering

Full Text:


Supplementary Files:


  • There are currently no refbacks.