摘要
在简介矿井瓦斯传感器关键技术的基础上,阐述了气体传感器漂移的抑制,包括传感器零点、灵敏度和非线性的自动调校技术及其研究进展,介绍了几种最新技术,涉及遗传算法、小波分解以及用DSP的实现方法,尤其是基于神经网络的传感器非线性自动调校方法,用传感器的输出和待测物理量的实际数值训练神经网络,以得到非线性校正用的逆模型。RBF神经网络的收敛速度、分类能力和逼近能力都比较好,是目前的研究热点。
Based on a brief introduction of the key technology of mine methane sensor,this paper mainly deals with the drift reduction of gas sensor,including the zero-adjustment,sensitivity correction and nonlinear compensation of methane sensor and their research progress.Several new techniques were introduced,such as genetic algorithm,wavelet decomposition and the implementation method using DSP.Especially the sensor′s nonlinear auto-calibration method using neural network was described.The RBF neural network was used as an inverse model that was trained to perform the mapping among the sensor′s readings and the actually sensed properties.With its converging speed,classification capability and approach capability,the RBF neural network has become a hot research area.
出处
《江苏工业学院学报》
2004年第2期61-64,共4页
Journal of Jiangsu Polytechnic University
基金
江苏省科技厅社会发展项目基金资助(BS2003012)
关键词
瓦斯传感器
自动调校
综述
methane sensor
auto-calibration
review
