Published May 16, 2026 | Version v2
Preprint Open

五维系统论驱动的轴承跨域统一健康评估——西安交大/美国/法国异构数据的全局理想体验证

Authors/Creators

Description

轴承健康评估对旋转机械预知性维护至关重要。传统方法依赖单一特征,难以刻画系统整体退化的协同规律。本文基于五维系统论,从边界、结构、储备、方向、强度五个维度构建轴承系统的全息映射,以乘积形式定义协同系数κ作为统一健康度量。针对西安交大、美国、法国三组异构数据集共48组轴承,建立理想体迭代归一化框架,实现跨数据集定量对比。统计分析与典型案例时序检验表明:早期退化阶段D-方向与κ的Pearson相关系数高达0.87~0.95,是系统退化的"先知维度";晚期退化阶段S-结构以r=0.88接管主导权,呈现明确的"维度主导权转移"(D→S)范式。双传感器配置显著增强κ曲线的稳定性。本文提炼出五种退化模式与八种典型案例,证明健康阈值0.8与崩解阈值0.1具有跨域普适性。进一步验证全局统一理想体的跨域有效性,为轴承健康管理及车路协同、空地一体化等跨领域应用提供统一评估框架。本文首次将五维系统论这一系统科学新框架应用于旋转机械健康评估的工程技术领域。

Abstract (English)

Rolling bearing health assessment is critical for predictive maintenance of rotating machinery. Traditional methods rely on single features and struggle to characterize the synergistic patterns of system-wide degradation. Based on Five-Dimensional Systems Theory (5ma), this paper constructs a holographic mapping of bearing systems across five dimensions—Boundary (B), Structure (S), Reserve (R), Direction (D), and Intensity (I)—and defines the synergy coefficient κ as the product of five-dimensional matching degrees. Using 48 bearings from three heterogeneous datasets (XJTU-SY China, NASA USA, FEMTO France), an iterative ideal-body normalization framework is established to enable quantitative cross-dataset comparison. Statistical analysis and case studies show that: (1) early-stage degradation is led by the D-Direction dimension (Pearson r = 0.87–0.95), the 'prophet dimension'; (2) late-stage degradation is dominated by S-Structure (r = 0.88), exhibiting a clear dimensional dominance transfer (D→S); (3) dual-sensor configurations significantly enhance κ stability; (4) health threshold 0.8 and collapse threshold 0.1 possess cross-domain universality. Global unified ideal-body validation further confirms cross-domain effectiveness, providing a unified assessment framework for bearing health management and cross-domain applications such as vehicle-road synergy and air-ground integration. This is the first engineering application of Five-Dimensional Systems Theory to rotating machinery health assessment, opening a new path for deep integration between systems science theory and engineering practice.

English version will be uploaded separately as an independent Zenodo record.

Files

appendix.zip

Files (216.1 MB)

Name Size Download all
md5:e91072270146acdede5f4b953922794b
206.9 MB Preview Download
md5:1ce17cd88a75c4e5602c93b4216fdf39
9.2 MB Preview Download