Privacy-aware and capacity-constrained scheduling of residential electric vehicle charging using action-driven optimization

Residential electric vehicle (EV) charging scheduling increasingly relies on fine-grained user-side information, such as arrival time, departure time, charging demand, and flexible charging preference. Although such information enables cost-effective and grid-friendly scheduling, it may also reveal sensitive mobility and household behavior patterns. Meanwhile, uncoordinated or abnormal charging requests can aggravate peak loads and threaten the operational security of residential distribution systems. To address these issues, this paper proposes a privacy-aware and security-constrained EV charging scheduling framework for residential communities. The framework models the charging process as an action-driven decision-making problem, where charging actions are optimized under user demand constraints, time-varying electricity prices, and grid capacity limits. Instead of relying on detailed mobility traces, the scheduler only uses compact charging request parameters required for decision making, thereby reducing the amount of user-side information exposed to the scheduler. A genetic algorithm is employed as an optimization solver for the resulting nonlinear and combinatorial problem. Experimental results under dynamic pricing and capacity-constrained residential scenarios show that the proposed method can reduce charging cost, flatten aggregate load, and mitigate peak charging risks while maintaining user charging requirements. The results demonstrate the potential of action-driven optimization for data-minimization-aware and operationally safe residential EV charging management.