Skip to content
Open access

Air Quality Prediction Using a Multi-strategy Improved Moss Gross Optimization Algorithm for Optimizing BiGRU Data Fusion

Sep 2026 · International Journal of Electric Power and Energy Studies · 0 citations · 20 references

TL;DR

A multi-strategy Improved Moss Growth Optimization (IMGO) algorithm designed to optimize a Bidirectional Gated Recurrent Unit (BiGRU) network, confirming that the proposed model delivers high precision and robust error reduction in air quality forecasting.

Abstract

The Air Quality Index (AQI) serves as a crucial indicator of dynamic trends in air pollution. However, achieving high prediction accuracy remains challenging due to the complexity of the atmospheric system, which is driven by intricate interactions between air pollutants and meteorological conditions. To address this and achieve precise air quality forecasting, this paper proposes a multi-strategy Improved Moss Growth Optimization (IMGO) algorithm designed to optimize a Bidirectional Gated Recurrent Unit (BiGRU) network. First, strategies including Fitness-Distance Balance (FDB), adaptive t-distribution, and Opposition-Based Learning (OBL) are integrated to enhance the optimization precision and convergence speed of the algorithm. Evaluations on six standard benchmark functions-supported by statistical results, convergence curves, and box plots-demonstrate that IMGO exhibits superior global optimization capabilities compared to the standard MGO. Second, IMGO is utilized to optimize the hyperparameters of the BiGRU model, significantly bolstering its stability. Finally, precise data fusion is accomplished leveraging the nonlinear mapping capability of the optimized BiGRU. The proposed IMGO-BiGRU model, alongside standard BiGRU and MGO-BiGRU, was applied to AQI prediction tasks. Experimental results indicate that IMGO-BiGRU achieves a high prediction accuracy of 0.982, with a Mean Absolute Error (MAE) of 9.710, a Mean Absolute Percentage Error (MAPE) of 8.107%, and a Root Mean Square Error (RMSE) of 17.660. These findings confirm that the proposed model delivers high precision and robust error reduction in air quality forecasting.

Read PDF

Similar papers

Conference Open access Sep 2026

A Hyperparameter-Optimized LSTM Machine Learning Method for Predicting Air Quality

The experimental results reveal that the XGBoost model outperforms the other models, while hyperparameter tuning further improves the effectiveness of both the Decision Tree and KNN models.

K. Venkatesh, A. Teja, Research, Guntur, India · 0 citations
Conference Open access 2026

Enhanced Air Quality Index Forecasting through Genetic Algorithm-Based Kernel Extreme Learning Machine

The proposed GA-KELM forecasting model is investigated through experiments using long-term data sets recorded by monitoring air pollution of a metropolitan city in China and exhibits higher prediction accuracy, smaller forecasting error, and better robustness compared to the existing models.

M. Prince, B. Rajalingam, D. Soundaravalli et al. · 0 citations
Sep 2026

Optimal machine learning techniques to predict air quality index

This work uses various ML algorithms to predict Air Quality Index (AQI), including ensemble methods such as bagging and boosting, and aims to identify the most efficient ML framework that is scalable and reusable for any city of choice.

M. Sumathi, Hemanth Babu Chava, P. Manjunath et al. · 0 citations
Sep 2026

An intelligent AQI prediction method using transfer learning and an IASO-optimized DCM-DESN under multi-source data: A case study of four Chinese cities

An intelligent AQI monitoring framework using transfer learning and an improved atomic search optimization (ASO) algorithm optimized deep echo state network with a dynamic activation mechanism (DCM-DESN) under multi-source data is proposed.

Yuan-Hao Du, Xiu-Li Geng, Sheng Cheng et al. · 0 citations
Open access Aug 2026

Adaptive Multi-Task Learning for Urban Air Quality Assessment

A Temporal-Aware Multi-Task Learning (TMTL-AQI) framework to assess urban air quality via structured data that outperforms single-task and baseline multi-task models with an accuracy of 0.7605 and an F1-score of 0.7422.

Iman Youssif Ibrahim, D. M. Ahmed · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.