Accurate sensing of shaft wind speed is the core link to realize the intelligent ventilation system, but at present, mine shafts are affected by airflow fluctuations caused by turbulence disturbance, and there are large errors in the wind speed monitoring results. In order to eliminate such errors, this paper proposes an adaptive and accurate sensing algorithm for shaft wind speed. The algorithm continuously optimizes the wind speed estimation and reduces the data noise by predicting and updating steps, combined with adaptive adjustment of the model error, so as to eliminate the influence of turbulence. Six types of shaft wind speeds, 0.15 m/s, 0.50 m/s, 2.00 m/s, 5.00 m/s, 10.00 m/s, and 15.00 m/s, were monitored continuously for 600 seconds, and the monitoring data were corrected using the algorithm studied in this paper. The analysis results show that the corrected data of this algorithm can be significantly close to the expected value compared with the monitoring data, and the accuracy has been improved by 73.56%, 65.04%, 56.24%, 61.38%, 71.04% and 69.63%, respectively. This algorithm provides accurate and reliable data support for the intelligent ventilation system and promotes the development of mine intelligence.