Scholarlink

<p>Water quality monitoring plays a critical role in safeguarding human health and environmental sustainability. However, existing machine learning models such as KNN, SVM, and CNN struggle with imbalanced and small-sample datasets, reducing their effectiveness for real-time water quality assessment. To overcome these limitations, this study introduces an innovative hybrid African Vulture Optimization Algorithm–Recurrent Neural Network (AVOA-RNN) framework. The novelty of the approach lies in three aspects: (i) the integration of AVOA with RNN to automatically tune hyper-parameters and select discriminative features, (ii) the incorporation of Synthetic Minority Oversampling Technique (SMOTE) to mitigate class imbalance and enhance minority-class recognition, and (iii) the evaluation on a newly collected Cauvery River water-quality dataset. Experimental results demonstrate that AVOA-RNN achieves 97% classification accuracy, outperforming CNN, LSTM, GA-RNN, and PSO-RNN baselines by 6–15%. These findings highlight the robustness, adaptability, and superior predictive power of the proposed framework for imbalanced water quality datasets.</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

A new machine learning technique for predicting river water quality using AVOA-RNN

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月13日

查看全文
  • Rajkumar Y.,
  • Karpagalakshmi R.C.,
  • Vellingiri J.,
  • Kalaivanan K.,
  • Devi Mani

摘要

Water quality monitoring plays a critical role in safeguarding human health and environmental sustainability. However, existing machine learning models such as KNN, SVM, and CNN struggle with imbalanced and small-sample datasets, reducing their effectiveness for real-time water quality assessment. To overcome these limitations, this study introduces an innovative hybrid African Vulture Optimization Algorithm–Recurrent Neural Network (AVOA-RNN) framework. The novelty of the approach lies in three aspects: (i) the integration of AVOA with RNN to automatically tune hyper-parameters and select discriminative features, (ii) the incorporation of Synthetic Minority Oversampling Technique (SMOTE) to mitigate class imbalance and enhance minority-class recognition, and (iii) the evaluation on a newly collected Cauvery River water-quality dataset. Experimental results demonstrate that AVOA-RNN achieves 97% classification accuracy, outperforming CNN, LSTM, GA-RNN, and PSO-RNN baselines by 6–15%. These findings highlight the robustness, adaptability, and superior predictive power of the proposed framework for imbalanced water quality datasets.