Cold thermal energy storage central air conditioning systems often fail to track demand response commands accurately, mainly due to pipeline transport delay and nonlinear electricity-to-cooling conversion in refrigeration units. To address this issue, this paper develops a nonlinear demand response control model with variable flow velocity and proposes a two-stage strategy combining steady-state scheduling and real-time control. In the scheduling stage, nonlinear electricity-cooling conversion and flow-velocity characteristics are linearized and discretized, and the mixed-integer nonlinear model is reformulated as a mixed-integer linear program via McCormick envelope relaxation. In the real-time stage, a sliding mode control law is designed for dual-objective tracking of demand response power and energy. Simulation results show that the proposed method improves demand response tracking accuracy, exploits system flexibility under comfort constraints, and enhances operational economy.

With the increasing penetration of renewable energy, cross-regional reserve sharing faces transmission congestion and high reserve procurement costs. This paper proposes a virtual reserve service framework based on spatially transferable loads in data centers. First, a data center migration capability assessment model and a cross-regional generator–data center joint reserve capacity optimization model are established to characterize the multi-constraint mechanism of virtual reserve provision. Second, a cross-regional virtual reserve consortium (CRVRC) mechanism is designed, incorporating joint quantity–price bidding and tiered pricing. Curve aggregation and tier-based allocation are used to coordinate reserve procurement between data center operators and generation operators. Finally, three representative case studies are constructed. The results show that the proposed method can integrate geographically distributed low-cost reserve resources, reduce reserve procurement costs, and support high-cost regions by coordinating data center task migration.