Сравнение методов оценивания риска ошибочного контроля качества природных вод

Abstract

A new concept for the quality control of natural waters was proposed. Currently, for this purpose a free method is used which involves the comparison of measured concentrations of each pollutant with the established standards, such as maximum permissible concentration ( MPC ), without specifying the probability of the fulfillment of C ≤ MPC or C > MPC type conditions. This control scheme corresponds to the concept of absolute precision of the analysis results and control. Following the modern theory of security it is necessary to get to the concept of acceptable risk, according to which the value of MPC must be supplemented by indicating the probability of finding the controlled parameter within the tolerance or the risk of going beyond its limits. In the mathematical formulation this is the task of constructing a confidence interval on the probability of performing a specified condition (tolerance interval). The current work explored different methods, applied to the problem of quality control of waters, by constructing parametric and nonparametric tolerance intervals, the first of which is independent of the law of distribution of a controlled parameter, while the second can be built under the normal law. We investigated the question of the amount of necessary measurements that should be performed to assess the water quality using each of these methods, and concluded that the parametric approach has substantial advantages. The paper also developed a method for separate confirmation of the requirements established for the mean concentration of the controlled parameter and its variance. It was shown that the required volume of test method occupies an intermediate position between the methods of parametric and nonparametric tolerance interval. To overcome the limitations in terms of normality of the relevant distribution law, a procedure was developed for estimating the required probabilities based on the mathematical apparatus of the order statistics that combines the advantages of both approaches. All of these studies were illustrated by computational examples that represent, in essence, the work practices of the practical application of the proposed approaches. Keywords: maximum allowable concentration, water quality control, non-parametric and parametric tolerance intervals, empirical probability distribution function, order statistics, conformity assessment (Russian) O.M. Rosenthal 1 , L.N. Alexandrowskja 2 1 of Water Problems of RAS, Gubkin ul. , 3, Moscow, 119991, Russian Federation 2 Moscow aviation Institute (national research University) (MAI), Volokolamskoe shosse, 4, 125993, Moscow, 125993, Russian Federation DOI: http://dx.doi.org/10.15826/analitika.2015.20.1.001