用径向基函数网络诊断氢发动机异常燃烧  

Abnormal Combustion Diagnosis of Hydrogen-fueled Engine Based on Radial Basis Function Neural Network

作　　者：段俊法[1] 郑俊强[1] 杨振中[1] 毕国进[1] 

机构地区：[1]华北水利水电学院,河南郑州450011

出　　处：《华北水利水电学院学报》2010年第2期55-58,共4页North China Institute of Water Conservancy and Hydroelectric Power

基　　金：国家自然科学基金项目(50322262);郑州市科技创新人才专项项目(096SYJH25086);郑州市技术研究与开发项目(0910SGYG23261-3)

摘　　要：消除异常燃烧是推进氢燃料发动机技术进步的关键之一,快速、有效、灵敏地获得异常燃烧信号是抑制异常燃烧的前提.分别对氢发动机正常燃烧和早燃压力信号进行时间序列采样,构造特征向量.将构造出的特征向量作为径向基函数网络的学习样本,利用径向基函数网络能逼近任意非线性函数的能力,对网络进行训练.结果表明:训练好的网络具有很好的泛化和学习能力,可以有效地诊断氢发动机的异常燃烧,并能灵敏地识别出异常燃烧故障的严重程度.Eliminating the abnormal combustion is one of a key technology which promotes hydrogen-fueled engine technology progress.Gaining effectively and sensitively the abnormal combustion signals is the basis to retrain the abnormal combustion.After sampling separately the normal combustion and preignition pressure signals of the hydrogen-fueled engine,then structuring the characteristic vector,the net was trained with the characteristic vector as its sample by means of the ability approaching to random nonlinear functions of Radial Basis Function Network.The result indicates that Neural Network after training has a good generalization and studying ability,and that Neural Network can effectively diagnose the abnormal combustion of hydrogen-fueled engine,also it can identify sensitively the graveness of the fault.

关 键 词：氢燃料发动机 异常燃烧 径向基函数网络 故障诊断 

分 类 号：TK463[动力工程及工程热物理—动力机械及工程]