Power Load Forecasting Assessment Method Based on Data Lineage Analysis

• Published in: IEEE Joint International Information Technology and Artificial Intelligence Conference (ITAIC ... ) (Online) Vol. 12; pp. 1422 - 1425
• Main Authors: Feng, Chaohan, Tao, Lei, Ye, Ruili, Xie, Lin, Wang, Yan, Feng, Qiong, Yu, Ling, Zhang, Zhoujie
• Format: Conference Proceeding
• Language: English
• Published: IEEE 23.05.2025
• Subjects: Accuracy; big data; data lineage; Forecasting; Load forecasting; load forecasting assessment; Logic; Neural networks; power system; Power system reliability; Predictive models; Training; Uncertainty; Vectors
• ISSN: 2693-2865
• DOI: 10.1109/ITAIC64559.2025.11163278

Electric load forecasting utilizes historical data of economic, social, meteorological, and other influencing factors that drive load development. It applies scientific tools and methods to conduct quantitative or qualitative analysis on the patterns of historical load data variations, thereby predicting future load conditions. However, due to the randomness and uncertainty of load fluctuations, different prediction subjects exhibit distinct characteristics, while various forecasting methods each have their own advantages and limitations. This necessitates adopting evaluation methods to assess forecasting accuracy.Data lineage emphasizes tracking the entire lifecycle of data, including its sources, processing logic, transmission paths, and assessment of influencing factors. This approach proves particularly suitable for evaluating load forecasting outcomes as it enables comprehensive tracing of data generation and transformation processes. By establishing clear causal relationships between raw data and prediction results, data lineage technology helps identify key factors affecting prediction deviations and optimizes model reliability.This paper proposes a method to evaluate power load forecasting results by analyzing actual load data and forecasted data, supporting power system decision-makers in making more rational decisions.