To enhance flue gas monitoring reliability under complex conditions, this study develops a novel data refinement framework integrating sensor array architecture with hybrid algorithmic processing. The proposed approach synergistically combines extended Kalman filtering and weighted averaging techniques to address nonlinear system challenges, achieving superior measurement precision and robustness. Benchmark tests demonstrate that our method outperforms traditional methods. When evaluated against the standard Kalman filter and its extended variant (EKF), the proposed algorithm achieves a 77.4% reduction in mean absolute error (MAE) and a 32.8% reduction in root mean square error (RMSE), as well as a 77.3% improvement in MAE and a 33.2% improvement in RMSE. Additionally, the algorithm outperforms mainstream alternatives such as particle filters and unscented Kalman filters in terms of metrics. Consequently, this study’s computational model demonstrates robust performance when deployed for IoT-based gas concentration measurement, particularly in industrial emission control and ecological sensing applications.