Abstract: Production, assembly, collision or wear will reduce the end accuracy of the anchor protection robot, and the deformation of the body caused by the load work will also affect the end accuracy.In order to reduce the anchor drilling error of the anchor guard robot, it can complete the tasks of downhole drilling, hole matching and support with high precision.This paper presents an error compensation method based on Sparrow algorithm improved BP neural network (SSA-BP algorithm).Firstly, the error model is built by the screw method, and the correctness of the error model is verified by the virtual prototype.Secondly, the error prediction model of the end pose is built to predict and compensate the error.Finally, through the comparison simulation of three compensation methods, SSA-BP, BP and PSO-BP, it is proved that the compensation accuracy of SSA-BP algorithm is higher and the stability is better.After experimental verification, the end error of the anchor guard robot is reduced to less than 10 mm, and the accuracy is improved by 80%.It can be seen that SSA-BP algorithm has excellent accuracy, superiority and feasibility in the error compensation of anchor protection robot, which provides efficient guarantee for safe mining.