DCAMamba: Mamba-based Rapid Response DC Arc Fault Detection

TL;DR

This work tackles the need for rapid and accurate DC arc fault detection in photovoltaic systems. It introduces DCAMamba, a state-space model–based framework powered by a hardware-aware parallel algorithm and enhanced by a Feature Amplification Strategy to improve discrimination of arc faults. Empirical results show 96.72% accuracy with a remarkably low inference time of 1.87 ms, outperforming CNN, RNN, RCNN, RNN-ATT, Transformer, and Mamba baselines. The approach demonstrates strong potential as a scalable backbone for industrial arc fault detection and signals a path toward open-source deployment after peer review.

Abstract

In electrical equipment, even minor contact issues can lead to arc faults. Traditional methods often struggle to balance the accuracy and rapid response required for effective arc fault detection. To address this challenge, we introduce DCAMamba, a novel framework for arc fault detection. Specifically, DCAMamba is built upon a state-space model (SSM) and utilizes a hardware-aware parallel algorithm, designed in a cyclic mode using the Mamba architecture. To meet the dual demands of high accuracy and fast response in arc fault detection, we have refined the original Mamba model and incorporated a Feature Amplification Strategy (FAS), a simple yet effective method that enhances the model's ability to interpret arc fault data. Experimental results show that DCAMamba, with FAS, achieves a 12$\%$ improvement in accuracy over the original Mamba, while maintaining an inference time of only 1.87 milliseconds. These results highlight the significant potential of DCAMamba as a future backbone for signal processing. Our code will be made open-source after peer review.

Figures (13)

• Figure 1: The Inference Time of our method takes only 1.87$ms$. In the same experimental environment, compared with the current mainstream methods, DCAMamba is even faster than simple CNN.
• Figure 2: Photovoltaic DC Series Arc Fault Experimental Platform.
• Figure 3: Sensor and motherboard design of the Arc Collection Device.
• Figure 4: The figure shows arc fault data obtained under different arc gap and arc speed conditions with a voltage of 100V. The first row (a1)(a2) represents the observations at 100V, with an arc gap of 0.1$mm$ and an arc speed of 0.1$mm/s$, where no arc occurs. The second row (b1)(b2) shows the results at 100V, with an arc gap of 0.2$mm$ and an arc speed of 0.8$mm/s$, where an arc fault occurs. The third row (c1)(c2) represents the observations at 100V, with an arc gap of 1.6$mm$ and an arc speed of 1.6$mm/s$, where an arc fault occurs.
• Figure 5: DCAMamba Pipeline: The data obtained from the experimental platform undergoes initial preprocessing to distinguish between fault and normal data. The processed data is then restructured by the FAS module. Next, the data is input into the DCAMamba model, where it is normalized before entering the model blocks. After computation in the Mamba Block, another normalization layer is applied, and the data is then passed to the classification head to produce the final result.