# PBZGNet: A Novel Defect Detection Network for Substation Equipment Based on Gradual Parallel Branch Architecture

## TL;DR

PBZGNet is a new deep learning network for detecting defects in substation equipment, offering better accuracy and efficiency than existing models.

## Contribution

PBZGNet introduces a gradual parallel-branch architecture with novel modules for improved multi-scale feature fusion and small target detection.

## Key findings

- PBZGNet outperforms YOLOv11 by 9.3% mAP@50 in lightweight variants.
- The full PBZGNet model achieves 7.3% higher mAP@50 than YOLO-SD, setting a new state of the art.
- The network's modules enhance detection of small targets and complex objects in cluttered substation scenes.

## Abstract

As power systems expand and grow smarter, the safe and steady operation of substation equipment has become a prerequisite for grid reliability. In cluttered substation scenes, however, existing deep learning detectors still struggle with small targets, multi-scale feature fusion, and precise localization. To overcome these limitations, we introduce PBZGNet, a defect-detection network that couples a gradual parallel-branch backbone, a zoom-fusion neck, and a global channel-recalibration module. First, BiCoreNet is embedded in the feature extractor: dual-core parallel paths, reversible residual links, and channel recalibration cooperate to mine fault-sensitive cues. Second, cross-scale ZFusion and Concat-CBFuse are dynamically merged so that no scale loses information; a hierarchical composite feature pyramid is then formed, strengthening the representation of both complex objects and tiny flaws. Third, an attention-guided decoupled detection head (ADHead) refines responses to obscured and minute defect patterns. Finally, within the Generalized Focal Loss framework, a quality rating scheme suppresses background interference while distribution regression sharpens the localization of small targets. Across all scales, PBZGNet clearly outperforms YOLOv11. Its lightweight variant, PBZGNet-n, attains 83.9% mAP@50 with only 2.91 M parameters and 7.7 GFLOPs—9.3% above YOLOv11-n. The full PBZGNet surpasses the current best substation model, YOLO-SD, by 7.3% mAP@50, setting a new state of the art (SOTA).