MTS多类判别研究

Study of Mahalanobis-Taguchi System in Multiple Recognition

MTS(马田系统 )是最新发展起来基于稳健思想提高模式识别等软件产品质量的综合计测法 ,该文通过采集各类别样本建立类别基准空间 ,分别在正交表上进行功能性评价使其优化后 ,将未知样本与各类别最优基准空间作用 ,根据马氏距离大小划分其归属 ,从而实现MTS的多类判别。将此方法应用于机车故障部位诊断 。

MTS (Mahalanobis Taguchi System) is an effective measurement method to improve the quality of software such as model recognition,but so far,it can only be used for classification of two models. In this paper, the theory and design model for MTS design in discrimination of multi class have been studied. After constructing several base spaces with samples, orthogonal array and R SN  ratio are used to estimate base space of each class and the best base spaces. Then, with the Mahalanobis distance of sample to each base space, the sample can be classified to the class it belongs to. Finally, as an example, the new approach has been applied to the fault diagnosis of engine and a satisfactory result has been obtained.