Scholarlink

<p>The article examines the metrological assurance of wastewater monitoring, specifically, at chemical industry facilities. Automated monitoring, which involves assessing the initial quality of effluents and the quality of their treatment prior to discharge into natural water bodies, helps to reduce water pollution. In order to improve the monitoring efficiency, it is necessary to address a&#xa0;priority innovation issue faced by water-intensive chemical industry facilities, i.e., to ensure reliable comparability of measurement results for various parameters of water composition and properties, as well as subsequent selective reduction in the concentration of particularly hazardous toxicants. It is shown that a&#xa0;need exists to improve the procedure for identifying pollutants (with their subsequent removal) that cause a&#xa0;hazardous increase in the informative indicators of anthropogenic water pollution. In hygiene, hydrochemistry, and ecology, such indicators include chemical and biochemical oxygen demand. The authors conducted a&#xa0;comparative analysis of methods for processing data (the results of measuring wastewater parameters at industrial facilities) since by selecting the most effective treatment method, the identification of particularly hazardous toxicants and a&#xa0;selective reduction in their concentration can be improved. On the example of studying the sewage wastewater of Kemerovo AZOT enterprise, the following processing methods were analyzed: predictive mathematics, conventional regression analysis, and neural network simulation. Neural networks proved to be highly effective, as they helped to identify the largest (maximum) number of causal relationships, as compared to other methods, including nonlinear relationships between pollutants and chemical and biochemical oxygen demand. According to the results of comparing the data processing methods for analysis of the causal relationships between the measured parameters of water composition and properties, it is recommended to use neural networks. The obtained results can be used to provide metrological support for environmental management, monitoring, and emission/discharge control (accounting) systems at nitrogen fertilizer facilities.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Metrological assurance of wastewater monitoring: application of neural networks in measurement data processing methods

  • 期刊论文

  • 发表日期：

    2026年5月6日

查看全文
  • Oleg M. Rozental,
  • Vladislav Kh. Fedotov

摘要

The article examines the metrological assurance of wastewater monitoring, specifically, at chemical industry facilities. Automated monitoring, which involves assessing the initial quality of effluents and the quality of their treatment prior to discharge into natural water bodies, helps to reduce water pollution. In order to improve the monitoring efficiency, it is necessary to address a priority innovation issue faced by water-intensive chemical industry facilities, i.e., to ensure reliable comparability of measurement results for various parameters of water composition and properties, as well as subsequent selective reduction in the concentration of particularly hazardous toxicants. It is shown that a need exists to improve the procedure for identifying pollutants (with their subsequent removal) that cause a hazardous increase in the informative indicators of anthropogenic water pollution. In hygiene, hydrochemistry, and ecology, such indicators include chemical and biochemical oxygen demand. The authors conducted a comparative analysis of methods for processing data (the results of measuring wastewater parameters at industrial facilities) since by selecting the most effective treatment method, the identification of particularly hazardous toxicants and a selective reduction in their concentration can be improved. On the example of studying the sewage wastewater of Kemerovo AZOT enterprise, the following processing methods were analyzed: predictive mathematics, conventional regression analysis, and neural network simulation. Neural networks proved to be highly effective, as they helped to identify the largest (maximum) number of causal relationships, as compared to other methods, including nonlinear relationships between pollutants and chemical and biochemical oxygen demand. According to the results of comparing the data processing methods for analysis of the causal relationships between the measured parameters of water composition and properties, it is recommended to use neural networks. The obtained results can be used to provide metrological support for environmental management, monitoring, and emission/discharge control (accounting) systems at nitrogen fertilizer facilities.