PI-STGNN model-driven risk assessment method for leapfrog tripping in photovoltaic power plants

Abstract

Unintended fault propagation in photovoltaic plants stems primarily from collector-line leapfrog tripping, a hazard characterized by multi-factor coupling, time-varying dynamics and hidden causal links that remain difficult to fully characterize under complex operating conditions. To tackle this challenge, this work presents a physics-informed spatio-temporal graph neural network (PI-STGNN) tailored for leapfrog tripping risk assessment: a hierarchical indicator set is constructed through systematic analysis of tripping mechanisms and protection coordination rules, a conditional diffusion model is deployed to generate multi-scenario augmented samples to alleviate real-world fault data scarcity, and physical constraints including protection time grading and electrical distance are explicitly embedded into the learning pipeline to strengthen the model’s capacity to capture risk evolution patterns and support accurate identification. For high-risk detection, the proposed approach achieves a precision of 0.92 and a recall of 0.91, delivering a notable reduction in missed detection of high-risk events.