基于贝叶斯方法的动平衡机动态测量不确定度分析

Bayesian-based dynamic measurement uncertainty analysis of dynamic balancing machines

以影响系数法为基础建立了动平衡测量模型,但是ISO-GUM提供的分析测量不确定度的方法无法解决其非线性,并且无法体现其测量系统全寿命周期内的动态性。针对这一问题,通过建立动平衡量与振动响应及影响系数的概率传递关系,利用Monte Carlo仿真方法计算动平衡量测量值的概率分布;同时,通过周期性的计量校准追踪影响系数的动态性,并利用贝叶斯方法抑制测量噪声对影响系数动态估计的干扰;最后,通过更新影响系数的分布将系统状态的动态性反映到测量不确定度的分析中。实验结果证明,该方法能够有效准确地追踪系统的动态性,并且能够在测量系统的全寿命周期内精确地反映测量系统的测量不确定度。

Based on influence coefficient method, the measurement model of dynamic balancing machines was established. However, the procedure provided by ISO-GUM for measurement uncertainty analysis cannot solve its nonlinearity and cannot reflect the dynamics of the measurement system during its whole lifecycle. Aiming at the above problems, the distribution propagation relationship between unbalance and vibration response as well as influence coefficients was set up, and then Monte Carlo simulation method was adopted to calculate the distribution of measurement values. Moreover, periodic metrological assessment was applied to trace the dynamics of influence coefficients, and Bayesian method was used to suppress the disturbance of measurement noises in the dynamic estimation of influence coefficients. Finally, through updating the distribution of influence coefficients, the dynamics of system states was reflected into the analysis of measurement uncertainty. Experiment results show that the proposed method can effectively and accurately trace the system dynamics, and can achieve accurate estimation of measurement uncertainty of measurement systems during its whole lifecycle.