Application of kalman filter for real-time methane concentration estimation in coal mines

Abstract

This study applies a one-dimensional Kalman Filter (KF) to improve methane concentration estimation using simulated noisy sensor data. The true concentration was modeled as a slowly varying random walk, and measurements were corrupted with Gaussian noise to mimic sensor inaccuracies. Implemented in MATLAB, the KF iteratively predicts and updates the concentration by optimally combining prior estimates with measurements. Results show that the KF effectively smooths fluctuations, closely tracks the true concentration, and outperforms raw sensor readings, quickly converging from initial offsets. Integrating KF-based estimation into methane monitoring enhances safety by providing stable, accurate realtime data, reducing false alarms, and supporting timely decision-making.