Abstract: The accidents caused by human factors in coal mine accidents are mainly caused by misoperation caused by fatigue and inattention of underground operating personnel. The existing personnel fatigue detection method based on physiological signals or eye images has the problems of complex implementation, poor adaptability, low accuracy and easy false negatives and false positives. In order to solve the above problems, a perception device of fatigue state of underground personnel based on head posture monitoring is designed. When the underground personnel is in a fatigue state, the head droop motion will occur, and the brain senses the head imbalance through the cochlea. In order to restore balance and keep awake, the brain will control the neck to carry out recovery head-raising motion, thus forming a periodic nodding motion. A nine-axis posture sensor is installed on the safety helmet to collect angular velocity, acceleration and magnetic field strength, and head posture data is obtained through data fusion. The head posture angle is calculated by using the quaternion method, and nodding motion is captured according to the head posture angle. When the proportion of nodding motion in unit time exceeds a threshold value, it is judged that the personnel is in a fatigue state, voice and light early warning occurs and an early warning signal is sent to a monitoring terminal on the ground through a wireless mode. The experimental results show that the device can accurately obtain the head posture angle, capture the fatigue characteristic motion, and judge whether the underground personnel is in a fatigue state effectively. The device has the characteristics of small volume, light weight, low power consumption and easy implementation, which can provide technical reference for operation fatigue monitoring in production operation practice.