Adaptive Data Augmentation for Power Load Synthesis Based on Training-Free Hidden Markov Models

View at publisher Summarise with VIRAT
Authors: Tiantong Qiao, Qingchun Hou, Ruisheng Diao, Bohan Qian, Shan Wu, Jing Zhang, Jianguo Qian
Cited by: 0
References: 10

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

With the escalating penetration of distributed renewable energy sources, long-term operational optimization and scheduling planning have become paramount for power system security and reliability. However, effective risk assessment and operational mode optimization require substantial load measurement or synthetic data, yet obtaining real data is often limited while existing synthetic data generation methods based on generative models require extensive real-world training datasets. This paper proposes a novel data augmentation framework leveraging Hidden Markov Models (HMM) for load fluctuation pattern recognition and synthetic load curve generation. The methodology employs a hierarchical autoregressive (AR) model for weekday pattern alignment, utilizes HMM to automatically identify distinct load volatility intervals, and applies adaptive noise injection tailored to different operational states to synthesize realistic load profiles. Comprehensive validation on two real-world datasets demonstrates that the proposed method exhibits excellent performance and practical application value for data support capabilities in evaluation systems considering multiple temporal dimensions.

Fields of Study & Subjects

Energy Load and Power Forecasting Smart Grid Energy Management Power System Optimization and Stability

Publication details

Journal / source
IEEE PES International Meeting (PES IM)
Pages
1–5
Publication date
18 January 2026
Language
EN