Modeling air pollution by integrating ANFIS and metaheuristic algorithms.

Area of Science:

• Environmental Science
• Artificial Intelligence
• Data Science

Background:

• Increasing urban populations and inadequate environmental policy implementation exacerbate air pollution.
• The adverse effects of air pollution on human health and ecosystems necessitate accurate prediction methods.
• Adaptive Neuro-Fuzzy Inference System (ANFIS) is a powerful AI technique for estimation but requires optimized training.

Purpose of the Study:

• To enhance air pollution modeling and prediction accuracy using ANFIS.
• To evaluate the effectiveness of metaheuristic optimization algorithms (GA, PSO, DE) in training ANFIS models.
• To compare the performance of metaheuristic-trained ANFIS with classical ANFIS for air quality forecasting.

Main Methods:

• ANFIS models were trained using Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and Differential Evolution (DE).
• Input data included various air pollutants (PM2.5, PM10, SO2, O3, NO2, CO) and meteorological parameters (wind speed, temperature, humidity, etc.).
• Daily air pollution levels in Istanbul were predicted and compared against classical ANFIS results.

Main Results:

• Metaheuristic-trained ANFIS models demonstrated superior performance in modeling and predicting air pollution compared to classical ANFIS.
• Optimization of ANFIS premise and consequent parameters using GA, PSO, and DE led to more effective estimations.
• The study successfully applied advanced AI techniques for accurate forecasting of key air quality indicators.

Conclusions:

• Metaheuristic optimization significantly improves the predictive capabilities of ANFIS for air pollution.
• Trained ANFIS approaches offer a more robust and accurate solution for environmental pollution modeling.
• This research highlights the potential of AI and optimization techniques in addressing critical environmental challenges.