基于遗传优化的PCA-SVM控制图模式识别

针对SVM和PCA-SVM进行质量控制图模式识别时泛化能力不足和识别精度不高的问题,提出一种基于遗传优化的PCA-SVM控制图模式识别方法。该方法的基本思想是首先基于特征子空间降维方法,运用PCA算法对原始特征样本进行主元分析,有效降低原始特征样本维数并突出聚类,提取各模式之间的主元特征;然后把此特征看成遗传算法中一组染色体,对支持向量机分类器核参数和惩罚因子进行二进制编码,通过对随机产生的一组染色体进行模式识别,并将此识别率作为遗传算法的适应度函数,通过选择、交叉和变异操作,对其参数进行自适应寻优;最后用优化的支持向量机分类器进行控制图模式识别。通过仿真进行验证,结果显示基于遗传优化的PCA-SVM分类器模型的控制图模式泛化能力强、识别精度高,可适用于生产现场质量控制。

PCA-SVM for control chart recognition of genetic optimization

Considering the problem that the precision and generalization are not ideal when recognize the basic patterns of quality control chart in PCA and PCA-SVM modeling, this paper proposed a control chart pattern recognition method based on genetic algorithm and PCA-SVM. The basic idea of the method was that, firstly, in view of the dimensionality reduction in feature space, used principal component analysis algorithm to lower the sample dimension, it also highlighted the clustering features. Then regarded the component characteristics as a chromosome which was then performed with binary code. It used a support vector machine classifier to recognized a random chromosome and considered recognition accuracy as the fitness function to evaluate the fitness of individual feature. By the operations of selection, crossover and mutation, with GA self-adaptive optimizing for penalty parameter and kernel parameter. Finally, it introduced the optimized SVM modeling to identify the control chart pattern. The simulation experimental results demonstrate that the proposed method has higher detection accuracy and stronger generalization ability than other methods, so it is more suitable for quality control in production field.