Long-distance running is an advantage of Chinese sports, but compared with the world level, there is still a big gap. Therefore, an advanced long-distance running training system is urgently needed to scientifically train our long-distance runners to change this situation. The purpose of this article is to study the long-distance running training system under inertial sensor network. According to the actual situation at home and abroad, a human gait analysis system based on inertial sensors is designed. Gait parameters are transformed into clinical medicine through related algorithms and software platforms. Experimental results show that although the step length calculated by the gait analysis system is different from the actual step length, the error value is small, kept below 3 cm, and the error percentage is less than 2%, which meets the accuracy requirements of gait analysis. This fully proves the feasibility of the zero-speed correction method in gait analysis.