Diffusion Model Cuts Power System Parameter Estimation Error by 58.6%
A new parameter estimation framework, the Joint Conditional Diffusion Model-based Inverse Problem Solver, addresses the ill-posed nature of inverse problems where different parameter combinations yield identical outputs. By leveraging diffusion model stochasticity, it generates candidate solutions capturing underlying parameter distributions conditioned on observations. Joint conditioning on multiple observations narrows posterior distributions of non-identifiable parameters. Applied to composite load model parameterization in dynamic power systems, the method achieves a 58.6% reduction in estimation error compared to single-condition models and accurately replicates system dynamic responses under various conditions. The research is detailed in arXiv:2411.10431.
Key facts
- Parameter estimation is a classical inverse problem often ill-posed due to non-uniqueness.
- The Joint Conditional Diffusion Model-based Inverse Problem Solver uses diffusion model stochasticity.
- Joint conditioning on multiple observations narrows posterior distributions of non-identifiable parameters.
- Applied to composite load model parameterization in dynamic power systems.
- Achieves 58.6% reduction in parameter estimation error compared to single-condition model.
- Accurately replicates system's dynamic responses under various conditions.
- Research published on arXiv with ID 2411.10431.
Entities
Institutions
- arXiv