In order to adapt to the complex situation faced by real-time processing of telemetry data, a real-time fusion method for telemetry data based on adaptive fault tolerance mechanism is proposed. Taking the full frame count as the main fusion criterion, it adapts to garbled code, frame loss, full frame disorder and other abnormal situations through the full frame caching mechanism, and combines key technologies such as full frame quality inspection, full frame quality optimization, time code assistance to judge the continuity of the full frame to achieve optimal quality and continuous stability of real-frame data flow in real time. The application results show that this method can make full and efficient use of multi-station telemetry data in real time and effectively improve the real-time fusion processing of multi-station telemetry data.