Scholarlink

Climate, socioeconomic, and environmental factors all have an impact on the high-quality water that Waste Water Treatment Plants (WWTPs) release for use in different applications. The concentration of important factors, such as Chemical Oxygen Demand (COD), is a crucial component in evaluating the quality of treated effluent (COD). Three different types of data were used to predict the COD: first, environmental data from the New Nicosia Waste Water Treatment Plant (NWWTP) M1; second, metrological data from the National Aeronautics and Space Administration (NASA) (at 2 m above Earth's surface) M2; third, a hybrid data M3 that combined the environmental and metrological data M1 and M2. In this work, XGboost, CatBoost and Random Forest (RF) models were employed to forecast Chemical Oxygen Demand. To understand how the models make predictions, an explainable AI models technique SHapley Additive exPlanations (SHAP) was utilize. The study demonstrates the impact of using a hybrid data combination for estimating COD in NWWTP the trustworthiness of the models employed.

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

Effects of Hybrid Metrological and Environmental Data for the Prediction of Chemical Oxygen Demand in Waste Water Treatment Plant Using Explainable AI Models

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Dilber Uzun Ozsahin,
  • Zubaida Said Ameen,
  • Berna Uzun,
  • Auwalu Saleh Mubarak,
  • Sani Isa Abba,
  • Salisu M. Lawal

摘要

Climate, socioeconomic, and environmental factors all have an impact on the high-quality water that Waste Water Treatment Plants (WWTPs) release for use in different applications. The concentration of important factors, such as Chemical Oxygen Demand (COD), is a crucial component in evaluating the quality of treated effluent (COD). Three different types of data were used to predict the COD: first, environmental data from the New Nicosia Waste Water Treatment Plant (NWWTP) M1; second, metrological data from the National Aeronautics and Space Administration (NASA) (at 2 m above Earth's surface) M2; third, a hybrid data M3 that combined the environmental and metrological data M1 and M2. In this work, XGboost, CatBoost and Random Forest (RF) models were employed to forecast Chemical Oxygen Demand. To understand how the models make predictions, an explainable AI models technique SHapley Additive exPlanations (SHAP) was utilize. The study demonstrates the impact of using a hybrid data combination for estimating COD in NWWTP the trustworthiness of the models employed.