Battery‐Aided Personalised Protection Strategy for Achieving Differential Privacy of Electricity Consumption Data

ABSTRACT With the continuous energy transition and digital industrial upgrades, the collection and analysis of electricity consumption data have become the foundation for supporting the intelligent power system operations and load‐side energy efficiency optimisation. However, this development raises significant concerns regarding user privacy and potential data leakage. To address the issue, this paper proposes a battery‐aided personalised differential privacy framework that generates physically realisable noise through battery operations, thereby overcoming grid stability limitations associated with conventional virtual noise injection methods. The proposed method enables adaptive privacy preservation through bounded noise probability functions with configurable privacy parameters and incorporates time‐of‐use pricing via a mean‐drift mechanism to enhance the economic viability of battery dispatch strategies. A multi‐objective sand cat swarm optimisation algorithm is employed to determine the optimal configuration of battery parameters. Numerical experiments demonstrate that the proposed method effectively defends against load forecasting attacks and significantly reduces electricity costs for users, offering a viable solution for balancing privacy protection and economic benefits in smart grids.