DPLib is an open-source, MATLAB-based benchmark library designed to advance distributed and decentralized power system analysis and optimization. Unlike centralized tools such as MATPOWER, DPLib provides a standard, reproducible library tailored for distributed studies, featuring:
20+ multi-region benchmark test cases of varying sizes.
A graph-based partitioning toolkit to decompose MATPOWER systems into electrically coherent regions.
Modular distributed OPF solvers:
ADMM-based DC-OPF solver (YALMIP)
ADMM-based AC-OPF solver (IPOPT)
The library generates standardized files and visualizations, enabling intuitive analysis, validation, and reproducible research in distributed optimization.
Access DPLib and resources on GitHub.
This work extends our research on temporal decomposition to accelerate multi-interval economic dispatch by leveraging machine learning for optimal subhorizon partitioning. Key Contributions:
Machine Learning-Aided Partitioning
Multiclass Load Profiles Decomposition
Click here to download the 24-bus, 118-bus, and 2006-bus test systems
This work addresses the synchronization bottleneck in distributed optimization, particularly under computational heterogeneity and communication delays. Key Contributions:
A learning-aided asynchronous ADMM
Momentum-Extrapolation Prediction-Correction
Online Anomaly Classification
Enhanced Robustness
Click here to download the 22-bus, 48-bus, and 118-bus test systems
This work introduces a temporal decomposition strategy to reduce computation time in SCUC. Key contributions include:
Multi-Subhorizon Decomposition
Linear Modeling
Accelerated Distributed Optimization
Tailored initialization strategy
Click here to download the 3-bus, 24-bus, and 118-bus test systems
This work proposes a prediction-correction-based asynchronous ADMM to overcome the synchronization bottleneck in iterative distributed optimization for OPF. Key Contributions:
Asynchronous Distributed Optimization
Momentum-Based Prediction
Correction Step for Stability
Click here to download the 22-bus, 48-bus, and 118-bus test systems
This work proposes a temporal decomposition strategy to enhance the scalability and computational efficiency of security-constrained economic dispatch under uncertainty. Key Contributions:
Parallel Subproblem Formulation
Intertemporal Dependency Modeling
Data-Driven Nonparametric Chance Constraints
Click here to download the 6-bus, 24-bus, and 472-bus test systems
This work proposes Accelerated Robust Analytical Target Cascading to improve the convergence and stability of distributed OPF algorithms. Key Contributions:
Balancing Coefficient for Objective Term Importance
Improved Convergence and Reduced Oscillation in ATC
Applicability to APP, ADMM, and Nesterov-based ADMM
Validation on DC and AC OPF Problems
Click here to download the 6-bus, 22-bus, 48-bus, and 118-bus test systems
This work presents a horizontal time decomposition strategy to accelerate security-constrained economic dispatch by reducing computational complexity and enabling distributed optimization. Key Contributions:
Multi-Subhorizon Time Decomposition
Overlapping Interval Modeling for Intertemporal Constraints
Accelerated Auxiliary Problem Principle Coordination
Initialization Strategy for Faster Convergence
Click here to download the 118-bus, 472-bus, and 4720-bus test systems
This work presents a nonparametric adaptive kernel density estimation algorithm to estimate probability density functions of power flow outputs in unbalanced distribution systems. Key Contributions:
Applicable to problems with unknown probability distributions
High accuracy with low computational cost
Models uncertainties of distributed generators
Provides complete statistical information for probabilistic power flow
Applicable to unbalanced power distribution systems
Click here to download the IEEE 13-bus and IEEE 37-bus test systems
This work presents a decentralized solution algorithm for network-constrained unit commitment in multiregional power systems, eliminating the need for a central coordinator. Key Contributions:
Fully decentralized and parallelized local unit commitment solutions
Bilevel optimization formulation for each control entity
KKT-based coordination between neighboring regions
Improved resilience to cyber-attacks and communication failures