In order to avoid the influence of changes in air humidity and uneven road surface at the location where the crane operates, the operating range calculated by the crane safety operation system and the actual minimum safe distance contour surface around the transmission wire are in contact, which affects the normal and safe operation of the substation. This paper proposes and implements a self-correction method for the safety height of substation cranes based on Kalman filtering. Firstly, the method integrates multi-source information, such as air temperature and humidity, attitude Angle between crane boom and horizontal plane and vertical plane, and crane boom elongation distance, and adopts multi-classification neural network intelligent algorithm to obtain the maximum safe operating height between the crane boom end and the high-voltage cable; then, combined with the Kalman filter algorithm to filter out the error data output by the laser ranging sensor, to achieve real-time correction and optimization of the vertical distance between the crane arm and the cable. The experimental results show that when the relative air humidity exceeds 80%, the safety alarm error rate is reduced by 12.76%, which effectively improves the reliability of the safe operation of the crane safety operation system when the air humidity exceeds 80%.