Verification of the Calculation Assumptions Applied to Solutions of the Acoustic Measurements Uncertainty

Abstract

The assessment of the uncertainty of measurement results, an essential problem in environmental acoustic investigations, is undertaken in the paper. An attention is drawn to the -usually omittedproblem of the verification of assumptions related to using the classic methods of the confidence intervals estimation, for the controlled measuring quantity.Especially the paper directs attention to the need of the verification of the assumption of the normal distribution of the measuring quantity set, being the base for the existing and binding procedures of the acoustic measurements assessment uncertainty. The essence of the undertaken problem concerns the binding legal and standard acts related to acoustic measurements and recommended in: 'Guide to the expression of uncertainty in measurement' (GUM) (OIML 1993), developed under the aegis of the International Bureau of Measures (BIPM). The model legitimacy of the hypothesis of the normal distribution of the measuring quantity set in acoustic measurements is discussed and supplemented by testing its likelihood on the environment acoustic results.The Jarque-Bery test based on skewness and flattening (curtosis) distribution measures was used for the analysis of results verifying the assumption. This test allows for the simultaneous analysis of the deviation from the normal distribution caused both by its skewness and flattening. The performed experiments concerned analyses of the distribution of sound levels: LD, LE, LN , LDWN, being the basic noise indicators in assessments of the environment acoustic hazards.

Analysis of the Accuracy of Uncertainty Noise Measurement

“…, n which describes the sample of the measured sound level, has a normal probability distribution. In practice, this may be not achieved especially for samples obtained from road or train noise measurements, which rarely have a normal distribution (Wszołek, Kłaczyński, 2006;Batko, Przysucha, 2013;Gałuszka, 2010;Batko, Bal, 2014). Moreover, the nature of the sample's probability distribution often remains unverified.…”