论文部分内容阅读
提出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方法在处理小样本量数据中的不足,且与蒙特卡洛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。
A method to evaluate the uncertainty of the dynamic characteristics of pressure sensor is proposed. First of all, using the shock tube dynamic calibration system to generate step pressure signal to stimulate the pressure sensor to obtain the sensor output signal. Secondly, based on empirical mode decomposition sensor output signal preprocessing method to reduce the impact of noise during dynamic calibration; Then, based on the input and output signals of the sensor, a mathematical model of the pressure sensor is established by using the adaptive least square method, and the dynamic characteristic parameters of the pressure sensor are obtained. Finally, according to the small sample characteristics of the dynamic characteristic parameter sequence obtained by the repeated calibration experiment, Self-help method to calculate parameters of extended uncertainty and relative uncertainty. The shock tube system was used to conduct repeated dynamic calibration experiments on the pressure sensor to calculate the uncertainty of the dynamic characteristic parameters in the time-frequency domain and to compare with the existing methods. The experimental results show that the proposed method can make up for the deficiencies of the Bessel method in processing small sample size data and the relative error between the method and the Monte Carlo method is less than 10%, which shows that the proposed method can effectively evaluate the dynamic characteristics of the pressure sensor Uncertainty of the parameters. The relative uncertainty of the dynamic characteristic parameters in the frequency domain is analyzed. The working frequency band and overshoot of the sensor are greatly affected by the noise, which provides an important basis for the improvement of the dynamic calibration experimental conditions.