基于Chebyshev网络和最佳逼近理论的CO_2浓度检测

CO_2 Concentration Detection Based on Chebyshev Neural Network and Best Approximation Theory

针对检测二氧化碳时因环境条件变化导致的检测精度降低、实时监测难、相似气体"交叉敏感"等问题,提出一种基于Chebyshev网络和最佳逼近理论的信息融合检测二氧化碳方法。该方法采用多个传感器采集多类信息,这些信息分别对应影响二氧化碳检测的物理量以及红外传感器测量数据,通过改进的Chebyshev网络进行信息融合,获得融合的检测结果。模拟试验结果表明:该方法明显地提高了检测二氧化碳的精度和鲁棒性。同时,采用信息融合技术,增强了气体检测的可控性。

In order to solve such CO2 detection problems as low precision,poor real-time monitoring,across sensitivity of a mixture of gases,this paper proposed an effective detecting method based on the best approximation theory and Chebyshev neural network.This method used several sensors to obtain multiple groups of sample data,which respectively correspond with physical quantity affecting CO2 detection and detecting data of infrared sensor.Then,it used modified Chebyshev neural network and achieved fused result.The experimental results show that the proposed way effectively improves detecting precision and robustness.At the same time,it availably enhances controllability of the carbon dioxide detection by adopting information fusion technology.