三维微纳米接触式探头系统动态测量不确定度建模与评定

Scientific evaluation of dynamic measurement uncertainty for the three-dimensional micro/nano touch probe is an unsolved problem in current measurement field, which cannot meet the requirement of fast and high precision measurement for three-dimensional geometric dimensions of MEMS devices. In this project, dynamic characteristics, the dynamic measurement uncertainty model and the uncertainty evaluation method of the probe are studied. The characteristics model of the probe system is established by using the transfer chain function method，which can reflect dynamic changes of the internal structure parameters. The sources and distributions of uncertainties in each module are analyzed and tested comprehensively. The propagation mechanisms of uncertainty factors are explored and then the dynamic measurement uncertainty model of the whole system is established. The uncertainty components caused by dynamic model and detecting parameters are evaluated by Bayesian dynamic model and combined to obtain the uncertainty of dynamic measurement results. Monte Carlo dynamic simulation method is used to verify the effectivity and reliability of the dynamic uncertainty model and the evaluation method for the probe. A novel method with strong versatility to evaluate dynamic measurement uncertainty for the probe is finally presented, which is not only beneficial to the enrichment and perfection of the theories and methods of dynamic measurement uncertainty evaluation, but also beneficial to the improvement and optimization of dynamic test performances of the probe, as well as promoting the development of MEMS related industries.

科学评定三维微纳米接触式探头的动态测量不确定度是当前测量领域未解决的难题，无法满足MEMS器件三维几何尺寸高精度、快速测量的需求。本项目研究探头的动态特性、动态测量不确定度模型以及动态测量不确定度评定方法。用传递链函数法建立能反映探头系统内部结构参数动态变化的特性模型；全面分析系统各模块不确定度因素及其分布特性，测试其分布参数，探索不确定度因素的传播机理，建立全系统动态测量不确定度模型；用贝叶斯动态模型评定动态特性模型、探测参数等引起的不确定度分量并合成动态测量不确定度；用蒙特卡罗动态模拟方法验证探头不确定度模型及评定方法的有效性与可靠性。最终形成具有较强普适性的探头动态测量不确定度建模与评定方法，进一步丰富与发展动态测量不确定度评定理论和方法，促进探头的性能改进和产业应用，推动MEMS相关产业的发展。

科学评定微纳米接触式探头系统的动态测量不确定度是当前测量领域未解决的难题，而动态特性不理想是接触式探头系统动态测量误差的重要来源，严重制约探头测量速度和精度的提升。为了提高探头系统的动态性能，分别基于Elman神经网络和贝叶斯反演算两种方法对探头系统进行了动态特性补偿，通过递推最小二乘法辨识出补偿前后探头系统的动态特性模型及动态特性参数。结果表明，两种补偿方法均显著提升了探头系统的动态特性，补偿后的最大振幅下降了74%，工作频带拓宽到112Hz，而且贝叶斯反演算补偿方法在实时性方面还优于Elman神经网络法。采用传递链误差与量值特性分析相结合的方法对探头系统的主要误差源，如信号感测误差、三维光学传感误差、信号处理误差、解耦标定误差、动态预行程误差、动态响应误差以及其他量值特性误差等进行了理论分析与建模，得到了探头全系统的不确定度评定模型，并基于贝叶斯统计法与蒙特卡洛法对探头系统进行了不确定度评定。将建模评定结果与仿真结果进行比较，验证了探头系统不确定模型与评定方法的有效性与可靠性，形成了具有一定适用性的探头系统动态测量不确定度建模与评定方法，进一步丰富与发展了测量系统不确定度评定理论与方法。