多核学习SVM在发动机故障诊断上的应用

Application of SVM Based on Multiple Kernels Learning in Engine Fault Diagnosis

为了从小样本的发动机性能参数故障数据中提取故障信息,准确而快速的识别故障数据,支持向量机具备的诸多优点使其逐步的应用到发动机的故障诊断中。目前,已有不少利用支持向量机进行发动机故障诊断的研究,但对核函数的选用大多使用单一核函数对数据进行训练,在没有先验的情况下,往往不容易选择出最优的核函数。通过对核函数的研究,提出将基于多核学习构建的组合核函数作为支持向量机新的映射函数,通过故障实例演示了多核学习SVM在发动机故障诊断上的应用,说明了组合核函数的构建方法和相比于单一核函数的优越性。

Support vector machine (SVM) is gradually used for engine fault diagnosis because of its outstanding advantages on extracting the features from small quantity samples of engine fault performance data and recognizing the fault quickly and accurately. Currently, many engine fault diagnosis studies are conducted by using SVM, but most of them select a certain kernel function to train their SVM, which is uneasy to choose the best one without any experience on the fault. Multiple kernels learning (MKL)is presented to build a combined kernel with weight coefficient as the new map function of SVM, meanwhile, a method is given to build a combined kernel and an example is used to demonstrate the application of the SVM based on MKL, which shows its superiorities compared with single kernel function SVM.