Szczegóły publikacji
Opis bibliograficzny
Probability distributions of one-day noise indicators in the process of the type A uncertainty evaluation of long-term noise indicators / Bartosz Przysucha, Agata Szeląg, Paweł PAWLIK // Applied Acoustics ; ISSN 0003-682X. — 2020 — vol. 161, art. no. 107158, s. 1–9. — Bibliogr. s. 9, Abstr.
Autorzy (3)
- Przysucha Bartosz
- Szeląg Agata
- AGHPawlik Paweł
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 126529 |
|---|---|
| Data dodania do BaDAP | 2020-01-13 |
| Tekst źródłowy | URL |
| DOI | 10.1016/j.apacoust.2019.107158 |
| Rok publikacji | 2020 |
| Typ publikacji | artykuł w czasopiśmie |
| Otwarty dostęp |  |
| Czasopismo/seria | Applied Acoustics |
Abstract
The estimation of long-term noise indicators using a small sample involves type A measurement uncertainty which is dependent upon the size of the sample and its probability distribution. If the distribution function of the data is different from normal, alternative methods of uncertainty assessment must be applied rather than the classical law of uncertainty propagation. This paper presents an analysis of the probability distributions of one-day noise indicators calculated from data provided by noise monitoring of the city of Gdańsk. The measurement data under analysis was comprehensive and diverse – the data included a four-year period and sixty-nine measurement stations; noise of different origins was recorded: traffic noise, railway noise, industrial noise and aircraft noise. The data set was analysed in terms of determining the forms of probability distributions. Only for 3% of the analysed one-day noise indicators, were the distribution functions confirmed to be normal. In this case, using the classical method (the law of uncertainty propagation) for determining type A measurement uncertainty may lead to the erroneous determination of the uncertainty interval. Following analysis of the data set, the possibility of modelling the distributions of one-day noise indicators with a mixture of two normal distributions was verified. Such an approach would significantly simplify uncertainty determination using the non-classical method based on probability distribution propagation. It was indicated that 94% of the analysed samples are characterised by a distribution that is a mixture of two normal distributions.