基于小波分析与支持向量机控制图混合模式识别

A Study on Pattern Recognition Based on WaveletAnalysis and Support Vector Machine for Hybrid of Control Chart

为提高控制图模式识别尤其是混合控制图模式识别的精度,在小波分析方法的基础上,提出各小波函数的不同分解层数与支持向量机分类器相结合的方法。首先,采用蒙特卡罗方法生成训练与测试数据集。其次,对数据集进行小波分解,提取其形状特征,选择合适的小波函数与分解层数。最后,运用小波变换提取数据集的重构特征,将小波重构特征输入训练好的支持向量机中进行模式识别,识别结果与小波分析BP神经网络模式识别方法进行对比。实验结果表明,Db4小波函数的三层分解与支持向量机相结合对混合控制图模式识别的精度较高,可有效应用于控制图模式识别。

In order to improve the accuracy of control chart pattern recognition,especially hybrid control chart pattern recognition,a method combining different decomposition levels of each wavelet function with support vector machine classifier is proposed based on wavelet analysis method.Firstly,Monte Carlo method is used to generate training and test data sets.Secondly,the data set is decomposed by wavelet,its shape features are extracted,and the appropriate wavelet function and decomposition levels are selected.Finally,the wavelet transform is used to extract the reconstruction features of the data set,and the wavelet reconstruction features are input into the trained support vector machine for pattern recognition.The recognition results are compared with the pattern recognition method of BP neural network based on wavelet analysis.The experimental results show that the combination of three-layer decomposition of Db4 wavelet function and support vector machine has high accuracy in pattern recognition of hybrid control charts,and can be effectively applied to pattern recognition of control charts.