Hybrid multilayer perceptron models optimized by evolutionary algorithms for urban air quality forecasting: a case study of Shiraz, Iran.

This research project provides implementation support and methodology guidance for developing hybrid multilayer perceptron (MLP) models optimized via evolutionary algorithms to forecast urban air quality, using Shiraz, Iran, as a primary case study. Given the complex, non-linear nature of atmospheric pollutant dispersion in metropolitan areas, standard neural network training algorithms often suffer from local minima convergence. To address this limitation, this work explores the integration of evolutionary optimization techniques—such as genetic algorithms or particle swarm optimization—to optimize the weights and biases of the MLP architecture. The proposed framework utilizes historical meteorological parameters and ambient air pollutant concentrations (such as PM2.5, PM10, NO2, and SO2) as input vectors. Through systematic simulation

and evaluation, the performance of the hybrid evolutionary-MLP models is benchmarked against conventional backpropagation neural networks and statistical forecasting methods. The project documentation supports the analysis of model accuracy using standard statistical metrics, including root mean square error and coefficient of determination. Ultimately, this research direction assists in structuring robust computational tools for municipal environmental planning, air quality index forecasting, and proactive public health management in highly populated urban centers.

Message: Hi FE, I need help with “Hybrid multilayer perceptron models optimized by evolutionary algorithms for urban air quality forecasting: a case study of Shiraz, Iran.” in “Environmental Engineering / Computational Modeling”