基于相关向量机模型的腐蚀声发射信号识别

相关向量机(RVM)模型的分类性能与其核函数参数的选择有密切关系。本文分别利用人工蜂群算法(ABC)、粒子群算法(PSO)和遗传算法(GA)寻找相关向量机模型的最优参数,对几种方法的寻优性能进行了对比。采用基于二叉树结构的一对多扩展方法,对二分类相关向量机模型进行了扩展,建立了四分类模型。基于该分类模型对罐底腐蚀声发射信号进行识别,将声发射特征参数和频域参数作为模型的输入参数,获得了较好的识别结果。

Corrosion Acoustic Emission Signal Recognition Based on Relevance Vector Machine Model

The classification performance of the RVM model and its associated kernel function parameter are closely related. This paper applies artificial bee colony algorithm （ABC）, particle swarm optimization （PSO） and genetic algorithm （GA） to find the optimal parameter of the RVM model, and the performance of these methods was compared. Based on the binary tree structure and one-against-all method, the binary-classification RVM model is extended to establish a four-classification model. The tank bottom corrosion acoustic emission signals were recognized using the established model. The characteristics parameters of the acoustic emission signal and the frequency-domain parameters were selected as the input parameters of the model, and a good recognition was obtained.