[2604.01725] LiteInception: A Lightweight and Interpretable Deep Learning Framework for General Aviation Fault Diagnosis

Computer Science > Artificial Intelligence arXiv:2604.01725 (cs) [Submitted on 2 Apr 2026] Title:LiteInception: A Lightweight and Interpretable Deep Learning Framework for General Aviation Fault Diagnosis Authors:Zhihuan Wei, Xinhang Chen, Danyang Han, Yang Hu, Jie Liu, Xuewen Miao, Guijiang Li View a PDF of the paper titled LiteInception: A Lightweight and Interpretable Deep Learning Framework for General Aviation Fault Diagnosis, by Zhihuan Wei and 6 other authors View PDF Abstract:General aviation fault diagnosis and efficient maintenance are critical to flight safety; however, deploying deep learning models on resource-constrained edge devices poses dual challenges in computational capacity and interpretability. This paper proposes LiteInception--a lightweight interpretable fault diagnosis framework designed for edge deployment. The framework adopts a two-stage cascaded architecture aligned with standard maintenance workflows: Stage 1 performs high-recall fault detection, and Stage 2 conducts fine-grained fault classification on anomalous samples, thereby decoupling optimization objectives and enabling on-demand allocation of computational resources. For model compression, a multi-method fusion strategy based on mutual information, gradient analysis, and SE attention weights is proposed to reduce the input sensor channels from 23 to 15, and a 1+1 branch LiteInception architecture is introduced that compresses InceptionTime parameters by 70%, accelerates CPU inference b...