Power transformers are essential components of modern electrical grids, making the accurate detection of appearance defects a vital task for maintaining system safety. The paper proposes a power transformer defect detection method based on a state space model and cross-domain feature fusion.
The proposed architecture begins with a multi-scale wavelet frequency enhancement module. This component uses three-layer Haar wavelet decomposition to extract frequency domain features. By employing a cross-domain gating mechanism, the system achieves adaptive complementary fusion of frequency-domain and spatial-domain features.
Second, a dual-branch state space modeling module is proposed. This branch utilizes the Mamba architecture for multi-directional selective scanning, which allows the model to capture global long-range dependencies effectively. This structure is combined with multi-scale convolution and ECA channel attention to extract local detail features, ensuring a robust enhancement of both global context and local information.
Finally, the method integrates a multi-scale linear attention module based on a dual positional encoding mechanism to enable efficient global context modeling. Experimental results validated the effectiveness of this integrated approach, which achieved an mAP50 score of 84.7%. This represents an improvement of 12.9 percentage points over the baseline method, validating the effectiveness of the proposed approach.
Subject of Research: Transformer Defect Detection
Article Title: A transformer defect detection method based on state space model and cross-domain feature fusion
Article References: Zhou, W., Wang, N., Li, Y., Lou, J., & Zheng, P. (2026). A transformer defect detection method based on state space model and cross-domain feature fusion. Scientific Reports. https://doi.org/10.1038/s41598-026-78578-z
Image Credits: AI Generated
DOI: 10.1038/s41598-026-78578-z
Keywords: Power Transformers, Defect Detection, State Space Model, Feature Fusion, Grid Safety, Computer Vision, Machine Learning, Haar Wavelet, Mamba Architecture, ECA Channel Attention, Predictive Maintenance, Infrastructure Security, Signal Processing, Multi-scale Analysis, Energy Systems
Cite Scienmag News
Blake Davidson. (November 19, 2026). New State Space Model Enhances Power Transformer Defect Detection Accuracy. Scienmag. https://scienmag.com/new-state-space-model-enhances-power-transformer-defect-detection-accuracy/
Blake Davidson. "New State Space Model Enhances Power Transformer Defect Detection Accuracy." Scienmag, 19 November 2026, https://scienmag.com/new-state-space-model-enhances-power-transformer-defect-detection-accuracy/. Accessed 19 November 2026.
Blake Davidson. "New State Space Model Enhances Power Transformer Defect Detection Accuracy." Scienmag. November 19, 2026. https://scienmag.com/new-state-space-model-enhances-power-transformer-defect-detection-accuracy/








