Comparison of ARIMA and ANN approaches in time-series predictions of traffic noise

• Garg, N ^[1] ; Sharma, M.K ^[1] ; Parmar, K.S ^[1] ; Soni, K ^[1] ; Singh, R.K ^[1] ; Maji, S. ^[1]
  1. [1] CSIR-National Physical Laboratory
• Localización: Noise Control Engineering Journal, ISSN 0736-2501, Vol. 64, Nº. 4, 2016, págs. 522-531
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • The paper analyzes the long-term noise monitoring data using autoregressive integrated moving averages (ARIMA) modeling technique and artificial neural networks (ANNs) methodology. Box-Jenkins ARIMA and ANN approach have been utilized to simulate daily equivalent LDay (06-22h) and LNight (22-06h) in A and C weightings for a period of 1 year. The forecasting performance is ascertained using the statistical tests. The work draws a comparison of time-series ARIMA and ANN approach for ascertaining their suitability for traffic noise modeling and forecasting. It is observed that the artificial neural network (ANN) models outperform the ARIMA models so developed. The pattern of ARIMAf orecasting models is directional and as such the time-series predictive model utilizing ANN approach has demonstrated superior performance over the ARIMA model.