The extensive integration of high-penetration renewable energy has highlighted the challenges associated with limited grid regulation resources. As emerging flexible resources in distribution networks

5G base stations can be aggregated to form a technical virtual power plant (TVPP) capable of responding to grid regulation commands. Given the necessity for privacy protection and rapid response during command disaggregation

this paper proposes a distributed grid command disaggregation method for TVPP of base stations based on privacy-security-ensured learning to optimization. First

a command disaggregation model that considers both distribution network security constraints and the base station's sleep modes is established. Then

by integrating Peaceman- Rachford splitting theory with Paillier cryptographic operations

a distributed privacy security co-optimization framework for the distribution network operator and TVPP is designed. Furthermore

based on learning to optimization theory

this paper develops a distributed command disaggregation method that employs neural network optimizers to approximate the optimal solutions of decomposed optimization sub-problems while ensuring solution feasibility through gauge mapping

thereby achieving fast distributed disaggregation of grid regulation commands by TVPP. Simulation results demonstrate the superior performance of the proposed method in terms of both privacy protection and solution speed.