The gravity energy storage system has high timeliness requirements for the phase of mass block placement. Addressing the inadequacy of existing grasping devices in terms of timeliness, a control method for a slope gravity energy storage system′s mass block grasping device based on deep neural networks is proposed. Firstly, the overall workflow of the slope gravity energy storage system and the requirements for the mass block grasping device are introduced. Secondly, a method for calculating the instantaneous acceleration of the mass block to determine its travel distance is proposed, and the error sources of this method are analyzed according factors affecting the travel distance of the mass block in the buffer platform. Based on these error sources, a dataset for the mass block′s travel distance is constructed. Finally, a deep neural network is introduced, and the network structure and parameters are determined through training. The Dropout mechanism is employed to enhance the generalization ability of the model, resulting in a predictive model for the mass block′s travel distance. The analysis indicates that the proposed control method can effectively meet the system′s requirements for the timeliness and accuracy of the placement phase. The predicted value of the mass block′s travel distance can be given within 50 ms, with a prediction error within ±0.1 m, Which verifies the feasibility of the method.