BOOST-RPF: Boosted Sequential Trees for Radial Power Flow

2026-03-23

AI Summary

This work addresses the poor scalability of traditional power flow solvers in distribution networks and the weak generalization of existing machine learning approaches by reframing radial distribution voltage prediction as a sequence learning task along root-to-leaf paths. The proposed architecture integrates path decomposition with residual modeling to align with the recursive physical nature of power flow computation. Built upon XGBoost, it introduces three residual formulations—absolute voltage, parent-node residual, and physics-informed residual—alongside edge-level supervision to achieve size independence and strong out-of-distribution robustness. Evaluated on the Kerber Dorfnetz and ENGAGE benchmarks, the method achieves state-of-the-art performance, with the Parent Residual variant consistently outperforming both analytical solvers and neural network baselines in accuracy and generalization while maintaining linear O(N) computational complexity.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Planning, Routing, and Scheduling: Planning/Scheduling and Learning
• Search and Optimization: Distributed Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Accurate power flow analysis is critical for modern distribution systems, yet classical solvers face scalability issues, and current machine learning models often struggle with generalization. We introduce BOOST-RPF, a novel method that reformulates voltage prediction from a global graph regression task into a sequential path-based learning problem. By decomposing radial networks into root-to-leaf paths, we leverage gradient-boosted decision trees (XGBoost) to model local voltage-drop regularities. We evaluate three architectural variants: Absolute Voltage, Parent Residual, and Physics-Informed Residual. This approach aligns the model architecture with the recursive physics of power flow, ensuring size-agnostic application and superior out-of-distribution robustness. Benchmarked against the Kerber Dorfnetz grid and the ENGAGE suite, BOOST-RPF achieves state-of-the-art results with its Parent Residual variant which consistently outperforms both analytical and neural baselines in standard accuracy and generalization tasks. While global Multi-Layer Perceptrons (MLPs) and Graph Neural Networks (GNNs) often suffer from performance degradation under topological shifts, BOOST-RPF maintains high precision across unseen feeders. Furthermore, the framework displays linear $O(N)$ computational scaling and significantly increased sample efficiency through per-edge supervision, offering a scalable and generalizable alternative for real-time distribution system operator (DSO) applications.

Problem

• power flow
• generalization
• distribution systems
• scalability
• topological shifts

Innovation

• BOOST-RPF
• sequential path-based learning
• gradient-boosted decision trees
• physics-informed residual
• distribution power flow