Abstract: For special coal mine environment resulting in a wide range of strong electromagnetic noise and interference, radio channel link quality is extremely unstable and may cause communication failure. The application of cognitive radio with wireless environment sensing intelligence and real-time best communication channel selection can guarantee normal communication of coal mine wireless system. The article proposes channel selection algorithm based on EWA learning, which allowes the system to learn coal mine radio environment communication channel characteristics online. By accumulating the history channel experience, it can quickly predict, select and change the current optimal communication channel, dynamic ensure the quality of communication links and finnally reduce system communication outage probability.