摘要
针对瓦斯传感器常见突发型故障,提出一种基于主元分析(PCA)和权重提升(WB)算法训练人工神经网络集成的瓦斯传感器故障诊断方法。利用PCA方法提取故障特征,得到的特征向量作为神经网络的训练样本;利用WB算法依次训练多个神经网络分类器;由集成神经网络得到待测样本的故障诊断结果。仿真实验表明:该方法对测试样本的识别正确率在98.5%以上,能够显著提高瓦斯传感器故障诊断的诊断精度和泛化能力。
Aiming at general abiupt faults of gas sensor, a method of gas sensor fault diagnosis is proposed based on principal component analysis （PCA）algorithm and artificial neural networks integrated via weight boosting （WB） algorithm. PCA method is used to exact fault feature, feature vectors are used as training samples, several ANN classifiers are trained in turn based on WB algorithm with these samples. Fault diagnosis result of sample under test is obtained by the artificial neural networks ensemble（ANNE） classifiers. Simulation results show that, identification correct rate of this method is more than 98.5 % ,it can significantly improve fault diagnosis precision and generalization performance of gas sensors.
出处
《传感器与微系统》
CSCD
2016年第9期33-35,38,共4页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(21265006)
江西省自然科学基金资助项目(20151BAB217006)
关键词
瓦斯传感器
故障诊断
主元分析
人工神经网络集成
权重提升
gas sensor
fault diagnosis
principal component analysis （PCA）
artificial neural networkensemble
weight boosting（WB）
