基于函数型连接神经网络的瓦斯传感器非线性校正  

Nonlinear correction of methane sensor based on functional link neural network

作　　者：郭全民[1] 王健[1] 

机构地区：[1]西安工业大学电子信息工程学院,陕西西安710032

出　　处：《传感器与微系统》2008年第10期5-7,共3页Transducer and Microsystem Technologies

摘　　要：通过对载体催化瓦斯传感器检测原理的分析,指出瓦斯体积分数与瓦斯传感器输出电压之间呈非线性关系,提出了应用函数型连接神经网络的强非线性逼近能力,且不依赖确定的数学模型的优点,建立非线性校正模型,消除瓦斯检测中的非线性误差。网络仿真结果与分段线性拟合曲线的比较表明:这种非线性校正模型结构简单、收敛速度快、逼近精度高。The nonlinear relation between methane volume fraction and catalytic methane sensor output voltage is indicated by analysis of the sensor detection principle. A nonlinear correction model based on functional link neural network （FLNN）, which is characterized by nonlinear approach ability and independent on accurate mathematical model, is proposed to eliminate the nonlinear errors in methane detection. The comparison between simulation result and pieeewise linear fitting curve shows that the model has advantageous of simple structure ,fast convergence, high approximation precision.

关 键 词：瓦斯传感器 函数型连接神经网络 非线性校正 

分 类 号：TM938.8[电气工程—电力电子与电力传动] TP212.6[自动化与计算机技术—检测技术与自动化装置]