Forecasting Dissolved Organic Carbon Levels Utilizing Metaheuristic Optimization with Artificial Intelligence Techniques
In recent years, there has been a noticeable surge in population, accompanied by the rapid development of industries, services, and agriculture within communities. This growth has intensified pressure on water resources, resulting in a simultaneous decline in both the quantity and quality of these v...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Bilijipub publisher
2024-12-01
|
| Series: | Advances in Engineering and Intelligence Systems |
| Subjects: | |
| Online Access: | https://aeis.bilijipub.com/article_212429_398e521dc4ca2277776ef4f720be9b8b.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In recent years, there has been a noticeable surge in population, accompanied by the rapid development of industries, services, and agriculture within communities. This growth has intensified pressure on water resources, resulting in a simultaneous decline in both the quantity and quality of these vital resources. Dissolved organic carbon (DOC) stands out as a pivotal factor in determining the quality of natural freshwater bodies. It serves as a crucial indicator of water pollution, offering insights into the presence of both natural and man-made organic pollutants. This study focuses on predicting the levels of dissolved organic carbon in South Florida utilizing artificial intelligence algorithms. Specifically, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Multilayer Perceptron (MLP), Radial Basis Function (RBF), Decision Tree (DT), and Support Vector Regression (SVR) algorithms were employed. The Sparrow Search Algorithm was utilized to fine-tune the hyperparameters of these algorithms. The effectiveness of the hybrid models generated was evaluated through correlation and error parameters, along with the number of iterations required to achieve the lowest error rate. Upon comprehensive review, the SVR-SSA hybrid model emerged as the most promising, exhibiting an R2 value of 0.997293, a Root Mean Square Error (RMSE) of 0.2906, and a Normalized Mean Square Error (NMSE) of 5×10-4 at the thousandth iteration in the train dataset. These values for the test data set are calculated as 0.999754, 0.1625, and 2×10-4, respectively. Hence, it is deemed the most optimum model for this research endeavour. |
|---|---|
| ISSN: | 2821-0263 |