Multi-Source Data State Estimation of Power System Based on Denoising Diffusion Implicit Model With Data Augmentation

Maosong Zhang

doi:10.1109/tase.2026.3730792

Multi-Source Data State Estimation of Power System Based on Denoising Diffusion Implicit Model With Data Augmentation

2026-09-04

0

PRE

AI

DOI：10.1109/tase.2026.3730792

Original

Original request for help

Save

Abstract

En 中文

Accurate state estimation is crucial for power system security, yet challenges persist in data scarcity and multi-source heterogeneity. This paper presents a novel framework that integrates the Denoising Diffusion Implicit Model (DDIM), a deep generative architecture, with advanced data augmentation and multi-source fusion techniques. The proposed physics-informed method synthesizes measurement data via iterative denoising, effectively expanding the data manifold while preserving critical system constraints. Distinguishing itself from conventional generative approaches, the framework introduces a reverse-optimized diffusion process that removes Markovian dependencies and enables adaptive control over generation iterations. This architectural innovation leads to notable improvements in both computational efficiency and sample fidelity. To further enhance robustness, a new multi-source metrological fusion algorithm is developed to reconcile temporal and spectral discrepancies across heterogeneous data sources, including PMU and SCADA measurements. This ensures reliable and coherent data integration for state estimation. Building upon this, an enhanced Weighted Least Squares Measurement (WLSM) algorithm is formulated to estimate state variables with higher accuracy. Comprehensive empirical evaluations, including benchmark comparisons with contemporary generative models and IEEE system case studies, demonstrate the framework’s superior performance in sample generation quality, estimation accuracy, and computational efficiency. These results underscore the method’s effectiveness in addressing the challenges of multi-source data integration, marking a significant advancement in deep learning-driven power system state estimation. Note to Practitioners—This work addresses the critical challenge of achieving accurate power grid state estimation when faced with incomplete or noisy data from SCADA, PMUs, and AMI systems. We present a practical solution that integrates a Denoising Diffusion Implicit Model (DDIM) with advanced multi-source data fusion techniques. Our approach generates realistic synthetic measurements to compensate for missing data while strictly adhering to grid physics constraints, intelligently reconciles timing and accuracy discrepancies between different data sources like slow SCADA and fast PMU measurements, and incorporates an enhanced Weighted Least Squares estimator that automatically prioritizes more reliable data. The main advantages of this work include significantly faster processing, automated bad data detection and correction to minimize manual intervention, and proven scalability as demonstrated in IEEE 118 and 57-bus system tests. While the solution requires initial offline training of the DDIM model and achieves optimal performance with PMU support, it offers grid operators a computationally efficient upgrade to existing energy management systems that significantly improves real-time monitoring accuracy without major infrastructure changes. This data-driven approach is particularly valuable for modern grids with increasing renewable integration and variable demand patterns.

Keywords:

Power system state estimation

data generation

multi-source data fusion

conditional denoising diffusion implicit modeling

Multi-Source Data State Estimation of Power System Based on Denoising Diffusion Implicit Model With Data Augmentation

Abstract

Journal

Organization

Cited Papers