Scholarlink

<p>Monitoring water quality is vital for protecting public health and ecosystems. Traditional Water Quality Index (WQI) assessment methods are labor-intensive and slow, limiting scalability. This study investigates the use of automated deep learning, specifically Google AutoML, for predicting WQI levels. The study utilizes water quality data collected from multiple monitoring stations along the Indus River (2018–2023), encompassing key physicochemical parameters such as pH, turbidity, TDS, COD, nitrate, and phosphate. A comparative analysis was conducted using water quality parameters, benchmarking AutoML against manually developed deep learning and machine learning models. Performance was evaluated using accuracy, precision, and interpretability. Results show that while conventional deep learning achieved slightly higher accuracy, Google AutoML delivered competitive results with significantly reduced development time and expertise requirements. These findings highlight AutoML as a practical and scalable tool for efficient water quality monitoring and support its integration into environmental management frameworks.</p> Graphical Abstract <p>The graphical abstract depicts the workflow for predicting water quality using deep learning. Data are first collected from monitoring stations equipped with sensors measuring pH, temperature, turbidity, and dissolved oxygen. These measurements are aggregated and organized in the data collection phase. The dataset is then processed through cleaning, normalization, and feature selection to ensure quality inputs for modeling. In the deep learning analysis stage, models are trained on historical and real-time data to identify patterns and anomalies related to water quality. The results are presented graphically, showing performance across different sample sites. Finally, the system predicts water quality in three categories—Good, Moderate, and polluted—providing actionable insights for policymakers and environmental agencies. This pipeline highlights the integration of sensor technologies with artificial intelligence for scalable and accurate real-time water quality assessment.</p> <p></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

Toward Intelligent Water Monitoring: Benchmarking AutoML and Deep Learning for WQI Prediction in the Indus River

  • 期刊论文

  • 发表日期：

    2025年12月1日

查看全文
  • Ejaz Ul Haq,
  • Nantawoot Inseeyong,
  • Jeerapong Laonamsai,
  • Kwanchai Pakoksung,
  • Mengzhen Xu,
  • Zhenxing Zhang,
  • Pavisorn Chuenchum

摘要

Monitoring water quality is vital for protecting public health and ecosystems. Traditional Water Quality Index (WQI) assessment methods are labor-intensive and slow, limiting scalability. This study investigates the use of automated deep learning, specifically Google AutoML, for predicting WQI levels. The study utilizes water quality data collected from multiple monitoring stations along the Indus River (2018–2023), encompassing key physicochemical parameters such as pH, turbidity, TDS, COD, nitrate, and phosphate. A comparative analysis was conducted using water quality parameters, benchmarking AutoML against manually developed deep learning and machine learning models. Performance was evaluated using accuracy, precision, and interpretability. Results show that while conventional deep learning achieved slightly higher accuracy, Google AutoML delivered competitive results with significantly reduced development time and expertise requirements. These findings highlight AutoML as a practical and scalable tool for efficient water quality monitoring and support its integration into environmental management frameworks.

Graphical Abstract

The graphical abstract depicts the workflow for predicting water quality using deep learning. Data are first collected from monitoring stations equipped with sensors measuring pH, temperature, turbidity, and dissolved oxygen. These measurements are aggregated and organized in the data collection phase. The dataset is then processed through cleaning, normalization, and feature selection to ensure quality inputs for modeling. In the deep learning analysis stage, models are trained on historical and real-time data to identify patterns and anomalies related to water quality. The results are presented graphically, showing performance across different sample sites. Finally, the system predicts water quality in three categories—Good, Moderate, and polluted—providing actionable insights for policymakers and environmental agencies. This pipeline highlights the integration of sensor technologies with artificial intelligence for scalable and accurate real-time water quality assessment.