一种冲击波压力传感器的准静态校准神经网络模型

发布时间：2018-03-15 00:29

本文选题：冲击波压力传感器　切入点：RBF神经网络　出处：《振动与冲击》2017年13期 　论文类型：期刊论文

【摘要】：由冲击波压力传感器准静态校准原理,间接比对式校准的精度取决于重锤的落高与压力拟合模型的精度,本文采用RBF神经网络建立了以落高为输入量、冲击波压力峰值为输出量的神经网络模型。选用典型标准压力传感器,在7~30 MPa量程范围开展校准实验;通过对测试样本进行分析,结果表明:该神经网络模型预测的最大相对误差不超过0.04%,比多项式拟合模型和指数拟合模型高一个数量级。落高与压力拟合模型引入的不确定度是构成冲击波压力传感器动态测量不确定度的一个重要分量,通过建立高精度的重锤落高与冲击波压力峰值神经网络拟合模型,为进一步提高冲击波压力传感器的测量精度奠定了基础。
[Abstract]:Based on the quasi-static calibration principle of shock wave pressure sensor, the accuracy of indirect comparison calibration depends on the precision of the falling height of the weight hammer and the precision of the pressure fitting model. In this paper, RBF neural network is used to establish the falling height as the input. A neural network model with the peak value of shock wave pressure as the output. The calibration experiment was carried out in the range of 730 MPa using a typical standard pressure sensor, and the test samples were analyzed. The results show that the maximum relative error predicted by the neural network model is less than 0.04, which is one order of magnitude higher than the polynomial fitting model and the exponential fitting model. The uncertainty introduced by the falling height and pressure fitting model is the formation of shock wave pressure. An important component of the dynamic measurement uncertainty of force sensors, A high precision neural network fitting model between the falling height of weight hammer and the peak value of shock wave pressure is established, which lays a foundation for further improving the measuring accuracy of shock wave pressure sensor.
【作者单位】：安徽工业大学电气与信息工程学院;南京理工大学机械工程学院;西安近代化学研究所;
【基金】：国家自然科学基金(11372143)
【分类号】：TP183;TP212

【参考文献】