Intelligent Power Grid Design Review via Active Perception-Enabled Multimodal Large Language Models

TL;DR

The paper tackles the challenge of automatically reviewing ultra-high-resolution power grid drawings, where traditional methods struggle with both computational load and holistic semantic reasoning. It introduces a three-stage framework powered by pre-trained Multimodal Large Language Models and advanced prompt engineering to perform global semantic understanding, region-specific high-resolution analysis, and confidence-aware design error diagnosis. Key contributions include a domain-specific semantic region proposal, fine-grained information acquisition with associated confidence scores, and a comprehensive decision module that outputs reliability assessments for detected issues. Preliminary results on a small in-house dataset indicate improved defect discovery and more reliable review judgments compared with passive MLLM inference, suggesting meaningful potential for scalable, intelligent review of critical infrastructure drawings.

Abstract

The intelligent review of power grid engineering design drawings is crucial for power system safety. However, current automated systems struggle with ultra-high-resolution drawings due to high computational demands, information loss, and a lack of holistic semantic understanding for design error identification. This paper proposes a novel three-stage framework for intelligent power grid drawing review, driven by pre-trained Multimodal Large Language Models (MLLMs) through advanced prompt engineering. Mimicking the human expert review process, the first stage leverages an MLLM for global semantic understanding to intelligently propose domain-specific semantic regions from a low-resolution overview. The second stage then performs high-resolution, fine-grained recognition within these proposed regions, acquiring detailed information with associated confidence scores. In the final stage, a comprehensive decision-making module integrates these confidence-aware results to accurately diagnose design errors and provide a reliability assessment. Preliminary results on real-world power grid drawings demonstrate our approach significantly enhances MLLM's ability to grasp macroscopic semantic information and pinpoint design errors, showing improved defect discovery accuracy and greater reliability in review judgments compared to traditional passive MLLM inference. This research offers a novel, prompt-driven paradigm for intelligent and reliable power grid drawing review.

Figures (7)

• Figure 1: High-level Overview of Our Proposed Three-Stage Framework for Intelligent Power Grid Drawing Review. The system processes an ultra-high-resolution drawing through semantic region proposal, fine-grained information acquisition, and comprehensive design error diagnosis.
• Figure 2: Overall System Architecture for Intelligent Power Grid Drawing Review. Our method leverages a three-stage framework with a comprehensive decision-making module.
• Figure 3: Simplified Prompt Template for the First Stage: Semantic Region Proposal. Illustrates the core instructions for the MLLM to identify domain-specific semantic regions from a global overview.
• Figure 4: Simplified Prompt Template for the Second Stage: Fine-grained Information Acquisition. Illustrates key instructions for extracting detailed elements and confidence scores from cropped high-resolution regions.
• Figure 5: Simplified Prompt Template for the Third Stage: Design Error Diagnosis. Illustrates the core instructions for synthesizing aggregated information to diagnose design errors and assess their reliability.