传感器静态误差校正方法研究报告.doc

中央民族大学学士学位论文 Bachelor Thesis of Central University of Nationalities 2010年4月25日 摘要 本文介绍了传感器非线性误差校正的原理和方法。它的主要方法有：分段线性化，数据拟合和神经网络。本文介绍了分段线性插值和曲线拟合的原理和方法，通过对光纤位移传感器位移、电压值的多项式拟合曲线的特点进行分析,发现该传感器曲线具有抛物线的特点所以采用通用多项式拟合。简要介绍了LabVIEW的结构和特点，并以光纤位移传感器为例，详细地介绍了LabVIEW环境下如何实现其实验数据曲线的拟合，为简化实验数据处理提供了一个良好的手段。 还介绍了2种应用颇为广泛的神经网络模型，BP及RBF神经网络的基本理论。阐述了2种算法的学习过程，其后简要地阐述了MATLAB神经网络工具箱设计BP和RBF网络的主要函数。以一受环境温度影响的光纤位移传感器为例，说明了具体实现方法和校正效果。仿真结果说明了BP及RBF网络具有很强的函数逼近能力。并分析BP及RBF网络结构和函数逼近能力的关系，得出网络的结构直接影响网络对函数的逼近能力和效果。 关键词：传感器；非线性；多项式拟合；神经网络；误差校正 Abstract The principles and method for correcting the nonlinear errors of the sensors are introduced. Its main methods are curve fitting method, the piecewise linear interpolation BP and RBF networks. It introduced principle and method of a new curve fitting method and the piecewise linear interpolation．The curve of a fiber-optic sensor possesses the characteristic of parabola by analysing the character of the curve obtained by normal polynomial fitting.So the new method of general polynomial fitting is used. With a simple introduction of the structure and characteristics of LabVIEW ,and by way of taking the fiber-optic sensor experiment as an example，how to realize the experimental data curve fitting through LabVIEW was illustrated，thus providing a favorable way to simplify the data processing. It introduced the basic theory of two popular ANN modules，which were BP and RBF networks．It demonstrated the learning procedure of each algorithm，and then briefly described main functions of their design in MATLAB neural network toolbox. To illustrate achieve method and correcting effect , an example of a fiber-optic sensor that affected by environment temperature is presented. approximation．The result shows that BP and RBF networks has a strong ability to approximate function．The paper also analyzes the relation between the structure of BP and RBF networks and precision of function approximate，educing that the structure BP feed forward n