A novel nowcasting (estimation) model based on an adaptive network neutrosophic hesitant fuzzy inference system (ANNHFIS): a case study of Istanbul

Abstract

Abstract Although biomass power plants are cleaner than fossil-fuel-based plants, they emit nitrogen dioxide (NO₂), which can degrade urban air quality and pose respiratory health risks. Therefore, reliable estimation (nowcasting) of NO₂ levels around these facilities is crucial for public health and air quality management. This study proposes an adaptive network-based neutrosophic hesitant fuzzy inference system optimized by particle swarm optimization (ANNHFIS-PSO) to estimate NO₂ concentrations near biomass plants in Istanbul. To our knowledge, this is the first adaptive neuro-fuzzy inference system (ANFIS)-based framework that incorporates neutrosophic hesitant fuzzy sets to represent environmental uncertainty. The proposed model integrates a neural network with neutrosophic hesitant fuzzy membership functions and employs a hybrid learning scheme that combines PSO-based global optimization with Adam-based fine-tuning to capture nonlinear relationships. Its performance was benchmarked against multilayer perceptron artificial neural network (MLP-ANN), ANFIS-PSO, grid-search-tuned ANFIS (ANFIS-GS), long short-term memory (LSTM) network and ANNHFIS-GS. Model accuracy was evaluated using metrics including root mean square error (RMSE) and coefficient of determination (R²). On the test dataset, ANNHFIS-PSO achieved an RMSE of 3.6488 µg/m³ and an R² of 0.8938, yielding the lowest RMSE and a high R² among the evaluated models. These results suggest that the proposed approach may support decision-making for air quality management near biomass plants.