Abstract During whole-body vibration (WBV), dynamic forces measured at the excitation-subject interface in directions other than the excitation axis, i.e. cross-axis response, are analysed using the principal component analysis (PCA) and virtual coherence techniques. The study applied these operations to the inline and cross-axis forces measured with twelve semi-supine human subjects exposed to longitudinal horizontal nominally random vibration between 0.25 and 20 Hz at root mean square acceleration levels of 0.125 ms−2 and 1.0 ms−2. The source identification is realised by a reversed path, aiming to identify relative contributions and correlations between the forces in response to a single axis excitation. The inline longitudinal and the cross-axis vertical forces were found to be correlated to each other from a low (e.g. 1–3 Hz) to a medium frequency range (e.g. 10–15 Hz). Above this range, where the forces were much reduced, the two forces tended to be independent in their contribution to the overall response. The singular vectors and virtual coherences were able to establish the degree of correlation in each of the frequency band identified. A signal processing framework is then proposed to take into account cross-axis responses for human vibration.

中文翻译：

---

全身振动过程中横轴视质量非线性的主成分分析

摘要 在全身振动(WBV) 过程中，使用主成分分析(PCA) 和虚拟相干技术分析在激励-主体界面处在激励轴以外方向上测量的动态力，即交叉轴响应。该研究将这些操作应用于 12 名半仰卧人体受试者在 0.125 ms-2 和 1.0 ms-2 均方根加速度水平下暴露于 0.25 和 20 Hz 之间的纵向水平名义随机振动时测量的内联和横轴力。源识别通过反向路径实现，旨在识别响应单轴激励的力之间的相对贡献和相关性。发现内联纵向力和横轴垂直力从低（例如 1–3 Hz) 到中频范围（例如 10–15 Hz）。在此范围之上，力大大减少，这两种力对整体响应的贡献往往是独立的。奇异向量和虚拟相干性能够在所识别的每个频带中建立相关程度。然后提出了一个信号处理框架，以考虑人体振动的跨轴响应。