Datasheet_YEH.csv
<p dir="ltr">This study presents advanced research on using Artificial Neural Networks (ANNs) to predict Chemical Oxygen Demand (COD) in wastewater treatment, specifically focusing on herbicide degradation. Here are the key aspects:</p><p dir="ltr">Research Obje...
Saved in:
| Main Author: | |
|---|---|
| Published: |
2024
|
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | <p dir="ltr">This study presents advanced research on using Artificial Neural Networks (ANNs) to predict Chemical Oxygen Demand (COD) in wastewater treatment, specifically focusing on herbicide degradation. Here are the key aspects:</p><p dir="ltr">Research Objective:</p><p dir="ltr">- To develop and evaluate ANN models for predicting COD removal during the degradation of commercial herbicides (Alazine and Gesaprim)</p><p dir="ltr">- To compare different backpropagation algorithms for training the neural networks</p><p dir="ltr">- To identify the most influential parameters affecting COD removal</p><p><br></p><p dir="ltr">Methodology:</p><p dir="ltr">1. Experimental Setup:</p><p dir="ltr">- Used a photochemical reactor with 250ml capacity</p><p dir="ltr">- Employed UV lamp (15W, 352nm) and ultrasonic probe (500W, 20kHz)</p><p dir="ltr">- Monitored six key variables:</p><p dir="ltr"> * Reaction time</p><p dir="ltr"> * pH</p><p dir="ltr"> * TiO₂ concentration</p><p dir="ltr"> * UV light intensity</p><p dir="ltr"> * Ultrasound frequency</p><p dir="ltr"> * Herbicide concentration</p><p><br></p><p dir="ltr">2. Neural Network Implementation:</p><p dir="ltr">- Tested five backpropagation algorithms:</p><p dir="ltr"> * Gradient Descent</p><p dir="ltr"> * Conjugate Gradient</p><p dir="ltr"> * Scaled Conjugate Gradient</p><p dir="ltr"> * Quasi-Newton</p><p dir="ltr"> * Levenberg-Marquardt</p><p dir="ltr">- Conducted 30 independent runs for each algorithm</p><p dir="ltr">- Split data: 70% training, 30% testing</p><p><br></p><p dir="ltr">Key Findings:</p><p dir="ltr">1. Algorithm Performance:</p><p dir="ltr">- Levenberg-Marquardt algorithm showed superior performance</p><p dir="ltr">- Achieved R² value of 0.9999</p><p dir="ltr">- Demonstrated lowest Mean Square Error (MSE)</p><p dir="ltr">- Showed statistical significance in performance difference compared to other algorithms</p><p><br></p><p dir="ltr">2. Parameter Influence:</p><p dir="ltr">- Reaction time was the most influential parameter (59.91% relative importance)</p><p dir="ltr">- Followed by:</p><p dir="ltr"> * Herbicide concentration</p><p dir="ltr"> * TiO₂ concentration</p><p dir="ltr"> * pH</p><p dir="ltr"> * Ultrasound frequency</p><p dir="ltr"> * UV light intensity</p><p><br></p><p dir="ltr">3. Model Accuracy:</p><p dir="ltr">- Successfully predicted COD values with high precision</p><p dir="ltr">- Showed strong correlation between experimental and predicted values</p><p dir="ltr">- Demonstrated robust performance across different operating conditions</p><p><br></p><p dir="ltr">Significance:</p><p dir="ltr">- Provides a reliable tool for real-time monitoring of wastewater treatment</p><p dir="ltr">- Offers potential for process optimization in industrial applications</p><p dir="ltr">- Contributes to more efficient herbicide removal from contaminated water</p><p dir="ltr">- Advances the field of environmental engineering through AI application</p><p><br></p><p dir="ltr">Our research represents a significant advancement in applying machine learning to environmental engineering, particularly in wastewater treatment optimization and herbicide degradation monitoring.</p><p><br></p> |
|---|