摘要
为了准确、有效地检测汽车尾气中各气体的质量分数,对传感器阵列和BP神经网络技术进行了研究,设计了一套汽车尾气检测系统。首先,根据汽车尾气成分选取4个相应传感器和一个温湿度传感器组成传感器阵列,搭建汽车尾气检测装置;其次,为了克服单一BP神经网络预测精度低,容易陷入局部极值的缺点,建立基于Adaboost算法和BP神经网络的集成神经网络模型;最后,利用集成神经网络模型对传感器阵列的响应信号进行回归分析。结果表明,集成神经网络模型预测的平均相对误差小于3%,能够有效处理汽车尾气的检测数据。
To test the mass fraction of gases in automobile exhaust accurately and effectively,an automobile exhaust detection system is designed by combining sensor array and BP neural network technologies in this study.An automobile exhaust detection device is built by madding a sensor array through the selection of four corresponding sensors and a temperature and humidity sensor according to the car's exhaust components. The single BP neural network can easily fall into the local extremum and has low prediction accuracy; thus,an integrated neural network model is established on the basis of the Adaboost algorithm and BP neural networks to overcome these disadvantages. This model is used for the regression analysis of experimental data. The results show that the average relative error predicted by the integrated neural network model is less than 3%.
出处
《环境工程学报》
CAS
CSCD
北大核心
2016年第4期1883-1887,共5页
Chinese Journal of Environmental Engineering
基金
国家自然科学基金资助项目(61471210)
浙江省宁波市科技局自然科学基金资助项目(2013A610002)