Related Experiment Video
Updated: Jul 13, 2026
11:53
Spatial Multiobjective Optimization of Agricultural Conservation Practices using a SWAT Model and an Evolutionary Algorithm
Published on: December 9, 2012
13.1K
Neuro-Particle Swarm Optimization Based In-Situ Prediction Model for Heavy Metals Concentration in Groundwater and
Kevin Lawrence M De Jesus1,2,3, Delia B Senoro1,2,3,4, Jennifer C Dela Cruz1,5
1School of Graduate Studies, Mapua University, Manila 1002, Philippines.
Toxics
|February 24, 2022
Summary
This study developed a neural network-particle swarm optimization (NN-PSO) model to predict heavy metal (HM) concentrations in water resources, offering a practical solution for limited monitoring data.
Area of Science:
- Environmental Science
- Water Resource Management
- Computational Chemistry
Background:
- Limited monitoring of heavy metal (HM) concentrations raises global environmental quality concerns.
- Elevated HM levels in water pose significant toxicological risks.
- Marinduque Island Province, Philippines, experienced mining disasters, necessitating water quality assessment.
Purpose of the Study:
- To predict heavy metal (HM) concentration in surface water (SW) and groundwater (GW) using in-situ physicochemical parameters.
- To develop and validate advanced prediction models for HM concentration.
- To address the challenge of limited HM concentration data in water resources.
Main Methods:
- Developed neural network-particle swarm optimization (NN-PSO) models for SW and GW.
- Utilized in-situ physicochemical parameters and historical HM data.
- Compared NN-PSO models against linear, support vector machine (SVM), and existing AI/deterministic models.
Main Results:
- NN-PSO models achieved near-ideal Mean Squared Error (MSE) values for both SW and GW.
- Kling-Gupta efficiency values consistently exceeded 0.95.
- NN-PSO models demonstrated superior predictive performance compared to linear and SVM models.
Conclusions:
- NN-PSO models are effective and practical for predicting HM concentration in water resources.
- The developed models provide a viable solution to supplement limited HM monitoring data.
- pH was identified as a significant factor influencing HM concentration via sensitivity analysis.
