Estimation of Water Quality Index using modern-day machine learning algorithms

Piyush Gupta,Pijush Samui, A. R. Quaff

Research Square (Research Square)(2023)

引用 0|浏览0
暂无评分
摘要
Abstract Many human-made activities currently pollute groundwater supplies, with mining operations playing a substantial role in this degradation. Water quality index (WQI) was calculated and forecasted for groundwater in gold mining sites of Kolar Gold Fields (KGF), Karnataka, using several water quality criteria. Aside from the difficulties in obtaining water quality monitoring for a specific location, artificial intelligence (AI) approaches have shown beneficial in consistently calculating target WQI based on optimum combination proportions. In the absence of available data, the most critical input parameters were discovered using three sophisticated deep learning models: convolution neural network (CNN), deep neural network (DNN), and recurrent neural network (RNN). The models were created utilizing the findings of seasonal monitoring experiments using various water quality metrics 80% of the experimental data was used to train the models, with the remainder used to validate the models. The best hyper-parameters for each model were trial-and-error selected; for CNN, DNN, and RNN, variable numbers of hidden layers, neurons, and training algorithms were used. Using parametric analyses on a simulated dataset, the trained models were validated. Mean absolute error (MAE), root mean square error (RMSE), determination coefficient (R2), Nash Sutcliffe efficiency (NSE), variance account factor (VAF), performance index (PI), Willmott's index of agreement (WI), mean bias error (MBE), and Root Relative Squared Error (RRSE) were used to assess the predictive abilities of proposed models. Furthermore, each model's strengths and weaknesses are examined. The researchers discovered that using CNN to do systematic calculations between water parameters and time series might be a useful tool for rapid water quality monitoring.
更多
查看译文
关键词
water quality index,machine learning,estimation,modern-day
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要