Defining Energy Indicators for Impact Identification on Aerospace Composites: A Structured Feature Selection Approach Guided by Domain Knowledge
定义航空复合材料冲击识别的能量指标:一种由领域知识指导的结构化特征选择方法
机构 * Department of Mechanics of Solids, Surfaces and Systems, Dynamics Based Maintenance (DBM) Group, University of Twente(固体力学、表面与系统部门,基于动态的维护(DBM)小组,特文特大学) ; Department of Aerospace Vehicles Integrity and Life Cycle Support (AVIL), Royal Netherlands Aerospace Centre(航空器完整性与生命周期支持(AVIL)部门,荷兰皇家航空研究中心)
专题命中 工作流自动化 :workflow(abstract);分类 cs.LG
AI总结 本研究提出一种由领域知识指导的结构化特征选择方法,构建能量指标作为全连接神经网络输入,使航空复合材料冲击识别的预测误差较传统方法降低三分之一,提升了可解释性与诊断置信度。
Comments revised framing from physics-informed to structured feature selection guided by domain knowledge, with greater emphasis on the link to impact dynamics; strengthened comparison with black-box CNN on traceability, interpretability, and model validation; results unchanged