基于特征聚集度的FCM-RSVM算法及其在人工焊点缺陷识别中的应用

An FCM-RSVM Algorithm Based on Feature Aggregation Degree and Its Application in Artificial Joints Defect Recognition

针对人工焊点缺陷识别方法进行研究,提出了一种基于特征聚集度的模糊C均值聚类(FCM)与松弛约束支持向量机(RSVM)联用的分类识别算法。在提取人工焊点特征向量的基础上,算法首先对样本特征数据进行模糊C均值聚类,依据样本隶属度函数计算不同特征的特征聚集度,并由特征聚集度指标改进RSVM算法中的松弛量参数,建立最终的分类器模型。实验结果表明:本文提出的算法建立了泛化能力更强的分类模型,能有效抑制噪声及模糊边界点对分类模型的影响,在人工焊点缺陷识别的应用中获得了满意的识别结果。

In order to improve the defect recognition of manual solder joints, this paper proposes a feature-aggregation-degree based combination algorithm of fuzzy C-means elustering（FCM） and relaxed support vector machine （RSVM）. Firstly, the characteristics of samples are extracted based on FCM algorithm and the feature aggregation degrees are calculated according to the different memberships. Then, the slack variable parameter of RSVM algorithm is repaired based on the feature aggregation degree such that the final classification model is established. The experiment results show that the proposed algorithm can effectively reduce the effect of noise or blur point on the classification model and build a stronger generalization classification model to improve the accuracy of defect recognition.