Robust Statistical Procedures in Statistical Analysis of Controllable Noise Indicators

The subject of the paper is statistical analysis of noise measurement results in control research of environment which is crucial for the process of identification, estimation, and interpretation of the characteristics of representative noise indicators. The proposed approach is based on robust methods of statistical analysis. In particular, the method of modeling measurement results with the use of autoregressive-moving average (ARMA) time series models is described. The subjects of identification, analysis, and connected matter of verification that occur during proposed mathematical formalization of considered problem are stated. Furthermore, the question of the validity and usability of normal distribution approximation, which is very common in modeling and estimation of the expected value of the equivalent sound level, is addressed. The illustration of the proposed methodology is represented by an example of the noise analysis from the permanent monitoring in the city of Kielce. The obtained results create the background for uncertainty of estimates of calculation concerning controlled noise indicators that describe the state of acoustic environmental threats.