压力传感器动态特性参数不确定度评定

压力传感器动态特性参数的不确定度是表征其动态测量性能的重要指标。提出了一种压力传感器动态特性参数的不确定度评定方法。首先，使用激波管动态校准系统产生阶跃压力信号激励压力传感器，得到传感器的输出信号；其次，采用基于经验模态分解（EMD）的传感器输出信号预处理方法，减小动态校准过程中噪声的影响；然后，根据传感器的输入输出信号，采用自适应最小二乘法建立压力传感器的数学模型，进而得到其时频域动态特性参数；最后，针对重复校准实验得到的动态特性参数序列的小样本特点，采用自助法计算参数的扩展不确定度和相对不确定度。采用激波管系统对压力传感器进行多次重复动态校准实验，计算时频域动态特性参数的不确定度，并与现有方法进行对比。实验结果表明：本文方法可以弥补贝塞尔法在处理小样本量数据中的不足，且与蒙特卡罗法的不确定度评定结果相对误差小于10%，说明本文方法可以有效地评定压力传感器动态特性参数的不确定度。分析时频域动态特性参数的相对不确定度得到传感器的工作频带和超调量受噪声的影响较大，为动态校准实验条件的改善提供了重要依据。

Uncertainty evaluation for dynamic characteristic parameters of pressure sensors

The uncertainty of dynamic characteristic parameters of pressure sensors is an important index to characterize its dynamic measurement performance. A method is proposed to evaluate the uncertainty of the sensor's dynamic characteristic parameters. Firstly, a shock tube dynamic calibration system is used to generate step pressure to excite the pressure sensor, and output signal of pressure sensor is obtained. Secondly, a preprocessing method based on empirical mode decomposition (EMD) is applied to reduce the influence of noise on output signals. Thirdly, an adaptive least squares method is performed to establish the mathematical model for pressure sensor based on its input and output signals, and the dynamic characteristic parameters in both time and frequency domains can be derived from the model. Finally, in consideration of the small sample feature of the parameter sequences in the repeated calibration experiments, the bootstrap method is applied to calculate the expanded uncertainty and relative uncertainty of these parameters. A set of dynamic calibration experiments for a pressure sensor are carried out with shock tube system. The uncertainties of dynamic characteristic parameters in both time and frequency domains are calculate and the results are compared with the existing methods. The experimental results show that the proposed method makes up the defect of the Bessel method in evaluating the sequence with small sample size. The relative errors of uncertainty results between the proposed method and Monte-Carlo method are less than 10%. It demonstrates that the proposed method works effectively in evaluating the uncertainty of pressure sensor's dynamic characteristic parameters. In addition, the analysis of the relative uncertainty evaluation results of the dynamic characteristic parameters in both time and frequency domains show that the work frequency band and overshoot are susceptible to the noises, and it can provide significant reference for the improvement of the dynamic calibration experiment conditions.