Information Flow Control in Off-Chain Components
本文研究了链下组件中的信息流控制问题，使用静态信息流控制技术来确保链上和链下组件间数据的完整性和保密性。
0 citationsRead paper
Selected work
Representative Papers
Data Compression for Time Series Modelling: A Case Study of Smart Grid Demand Forecasting
To address storage and transmission bottlenecks associated with high-frequency time-series data in smart grids, this study systematically investigates the impact of wavelet-based compression on load forecasting accuracy. We employ biorthogonal discrete wavelet transform (DWT) for multi-level lossy compression and—novelly—quantify the relationship between compression ratio and normalized mutual information (NMI) to assess information preservation. Robustness across forecasting models is evaluated using ordinary least squares (OLS), XGBoost, and the temporal diffusion encoder (TiDE), leveraging real-world data from the Hirtshals seawater supply system in Denmark. Results demonstrate that XGBoost exhibits exceptional robustness to compression-induced distortion: it maintains stable forecasting performance even at an extreme compression ratio of 99.9%, challenging the conventional assumption that compression inevitably degrades prediction accuracy. This work establishes a new paradigm for intelligent energy systems—one that reconciles data lightweighting with predictive reliability.
0 citationsRead paper
Recent publications
Latest Papers
Information Flow Control in Off-Chain Components
本文研究了链下组件中的信息流控制问题，使用静态信息流控制技术来确保链上和链下组件间数据的完整性和保密性。
0 citationsRead paper
Data Compression for Time Series Modelling: A Case Study of Smart Grid Demand Forecasting
To address storage and transmission bottlenecks associated with high-frequency time-series data in smart grids, this study systematically investigates the impact of wavelet-based compression on load forecasting accuracy. We employ biorthogonal discrete wavelet transform (DWT) for multi-level lossy compression and—novelly—quantify the relationship between compression ratio and normalized mutual information (NMI) to assess information preservation. Robustness across forecasting models is evaluated using ordinary least squares (OLS), XGBoost, and the temporal diffusion encoder (TiDE), leveraging real-world data from the Hirtshals seawater supply system in Denmark. Results demonstrate that XGBoost exhibits exceptional robustness to compression-induced distortion: it maintains stable forecasting performance even at an extreme compression ratio of 99.9%, challenging the conventional assumption that compression inevitably degrades prediction accuracy. This work establishes a new paradigm for intelligent energy systems—one that reconciles data lightweighting with predictive reliability.
0 citationsRead paper