Research on reliability from wastewater online monitoring data for CODcr with primary test method by CSS-bias-correction of Deming regression analysis

Abstract

Abstract There is a large number of online automatic devices, as the process stream analyzer undertaken testing in China, for their superiority compared with the laboratory technology. However, it has to be admitted that the online system, belonging to a non-standard, shall be paid more attention for its reliability. In this paper, for the chemical oxygen demand-dichromate method (CODcr) analysis in wastewater pollution sources using the online system (as X-method) [1–6] in comparison with primary test method (PTM as Y-method) [7], we, via Deming regression [8] of variable error at levels, fit with centered squares sum (CSS) correction between both methods, first unbiased (CSS0), and then constant bias (CSS1) or linear bias (CSS2). F and t, as well as χ2 distribution test are subsequently followed by for the selected CSS. To ensure residuals from the selected CSS under independent identical distribution (i.i.d) condition [9–15], Anderson Darling (AD) hypothesis test [16–19] is strongly recommended for i.i.d condition. All cumulative effects brought by the corrected online system under top-down site precision, are incorporated into the estimated value of the extended uncertainty to the maximum extent and avoided the complicated correlation, which is very helpful to improve the quality level of the online system.