在对气体进行长期在线监测的场合,所要解决的关键问题之一就是气体传感器特性漂移的 抑制。传感器特性漂移会给气体的测量和识别带来误差。对此,本文提出一种动态在线标定法。该方法能够对传感器漂移故障做出判定,同时可实现测量误差的修正。文中对该方法进行了详细的描述和论证。计算机仿真结果表明了该方法的有效性。 关键词: 气体传感器; 神经网络预测器; 混合气体识别 Abstract : The drift should be addressed as one of the most serious impairments for gas sensor. When drift exists , the gas measurement or identification result may cause error. To solve this , a dy2 namic on2line calibration method is proposed in this paper. By this method the sensor drift can be detected accurately and the measurement error caused by sensor drift may be corrected at the same time. Detailed descriptions of this method are given. The simulation results show that the method is valid. Key words : Gas sensor ; Neural network predictor ; Mixed gas identification