Research on Zero-Knowledge Proof–Integrated Privacy-Preserving Data-Cleaning Methods for Power-Grid Dispatch

To reconcile the trade-off between data quality assurance and privacy preservation in power grid time-series dispatch data, this paper presents a privacy-preserving data sanitization framework based on Zero-Knowledge Succinct Non-interactive Argument of Knowledge (zk-SNARK). The proposed framework integrates algebraic constraints and sliding-window statistics to enable local anomaly detection and missing-value imputation under zero-knowledge commitments. A non-interactive proof scheme is subsequently employed on the verification side to efficiently validate compliance with sanitization rules. Experimental results demonstrate that the method preserves data confidentiality while maintaining high accuracy in anomaly detection and imputation. Furthermore, the end-to-end processing latency satisfies the real-time requirements of power grid dispatch systems. Overall, this work offers a practical and effective solution for balancing privacy protection with data quality management in modern power systems.