Real-world aircraft health diagnosis requires balancing accuracy with computational constraints under extreme class imbalance and environmental uncertainty. This paper presents an engineering application of heterogeneous task decomposition for deployable intelligent fault diagnosis. The proposed Long-Micro Scale Diagnostician (LMSD) explicitly decouples global anomaly detection (full-sequence attention) from micro-scale fault classification (restricted receptive fields), resolving the receptive field paradox while minimizing training overhead. A knowledge distillation-based interpretability module provides physically traceable explanations for safety-critical validation. Experiments on the public National General Aviation Flight Information Database (NGAFID) dataset (28,935 flights, 36 categories) demonstrate 4-8% improvement in safety-critical metrics (MCWPM) with 4.2 times training acceleration and 46\% model compression compared to end-to-end baselines, substantiating deployability in resource-constrained aviation environments.

翻译：真实场景下的飞机健康诊断需要在极端类别不平衡和环境不确定性条件下，平衡诊断精度与计算资源约束。本文提出面向可部署智能故障诊断的异构任务分解工程应用方法。所提出的长微尺度诊断器（LMSD）通过显式解耦全局异常检测（全序列注意力机制）与微观尺度故障分类（受限感受野），在最小化训练开销的同时解决感受野悖论。基于知识蒸馏的可解释性模块为安全关键验证提供具有物理可追溯性的解释。在美国国家通用航空飞行信息数据库（NGAFID）公开数据集（28,935个航班，36个类别）上的实验表明：与端到端基线方法相比，本方法在安全关键指标（MCWPM）上提升4-8%，训练速度提升4.2倍，模型压缩率达46%，验证了其在资源受限航空环境中的可部署性。