较小样本动态声发射信号多元统计分析技术

Analysis of Dynamic Acoustic Emission Signals Using Multivariate Statistical Technique for Smaller Dataset

利用多传感信息集成系统,以两组平均年龄对应的受试对象往复运动过程中获取的动态声发射信号和角度信号为对象,研究了适用于较小样本的动态声发射信号多元统计分析技术。通过同步记录的角度信号,将往复运动分解为若干个独立运动周期和运动过程;利用累计概率分布,选取具备较显著差异的特征;结合多元统计技术,减小数据量,建立动态声发射信号的可视化模型,证实了使用较小样本声发射信号实现膝盖骨关节诊断的可行性。

A potential multivariate statistical based acoustic signal analysis and processing technique is presented for dynamic knee assessment for the smaller dataset.By using the integrated data acquisition system developed by the authors from two age-matched elder groups,the dynamic acoustic and the corresponding joint angle signals emitted from the consecutive knee movements are acquired.Consecutive movement cycles are isolated into individual for further analysis,and the cumulative probability distribution is employed for feature selection.Multivariate statistic methodologies are employed to derive the acoustic emission based joint profiles and to create the visual effect among the healthy and osteoarthritic groups,as well as to create the cluster evidence to demonstrate the feasibility of diagnosing knee osteoarthritis using the dynamic acoustics emission.The research findings show not only the potentials for simplifying the data acquisition protocol,but also the discovery of movement significancy.