基于神经网络的传感器静态特性曲线的拟合方法

传感器的静态特性是指被测量的值处于稳定状态时传感器的输入输出关系。随着传感器技术的发展，各种材料的新型传感器不断涌现，但是传感器静态特性的非线性和离散性往往无法避免，能否克服上述缺点已经成为传感器进入工业应用以及达到测量精度要求的主要问题。多层前向神经网络的误差逆传播算法（简称ＢＰ算法）的输入输出关系实际上是一种映射关系，适于解决上述非线性映射问题。利用神经网络处理传感器静态特性的非线性关系可以获得高度逼近实际静态特性的拟合曲线。

Method of Fitting Static Characteristic Curve for Sensor Based on Neural Network

The static characteristic of sensor means a relationship between the input and the output of the sensor when the value being measured is stable. With the development of sensor technology,new-type sensors based on various materials come forth, while the nonlinearity and dispersion of sensor′s static characteristic are unavoidable. It has become a primary problem that influences its industrial application and metrical precision. The I/O relationship of back propagation algorithm (BP algorithm) for Feed-Forward Multi-layered Neural Network is a mapping relationship, which can resolve the above nonlinear problem. The fitted curve of actual static characteristic for sensor can be gained by using BP algorithm.