恒温瓦斯检测输出非线性的动态调校

Dynamic Correction of Nonlinear Characteristic of Thermostatic Methane Detection with Catalytic Sensor

恒温瓦斯检测是一种使催化传感器在检测瓦斯气体时 ,始终处于恒温状态的新型检测方法 ,使瓦斯检测的稳定性 ,响应速度及传感器寿命等指标极大优于传统电桥检测方法 ,检测瓦斯的浓度范围也扩大了一倍 ,但输出特性呈现较严重的非线性 ,影响了瓦斯检测的准确性。本文在分析恒温瓦斯检测输出特性的基础上 ,采用神经网络自学习功能 ,对恒温瓦斯检测信号进行动态线性调校 ,使瓦斯检测获得较高的精度。

The thermostatic methane detection is a kind of new measuring technique, witch keeps the catalytic sensor in unchanged temperature by an automatic control circuit and obtains the better properties. Although the new measuring technique has made the response faster, the stability higher, the measuring range wider and the sensor life longer than present technique, its nonlinear characteristic has decreased the accuracy of the methane detection. With the variance parameter of the sensor, dynamic correction of nonlinear characteristic is very necessary for the accurate detection. A kind of neuron-network model, Adaline model, has been introduced in the work. The function of intelligent learning in neuron-network is very useful for the correction of the nonlinear characteristic of the sensor and has much improved the accuracy of methane detection.