摘要
提出了一种基于粒子群优化(particle swarm optimization,PSO)的特征提取算法,该算法以K-NN(nearest neighbor)分类正确率为评价准则,应用粒子群优化算法寻找使提取特征的K-NN分类正确率最大的转换矩阵,从而实现特征的提取。算法的特点是结构简单灵活,对数据的分布特征不敏感,适合于对模拟电路的故障特征进行提取。电路故障诊断示例证明了该特征提取算法在模拟电路故障诊断中的有效性。
A feature extraction algorithm based on PSO （particle swarm optimization） is proposed. K-NN （nearest neighbor） classification accuracy is used as the optimization criteria and PSO is used to search transformation matrix that maximize the K-NN classification accuracy of extracted features. The algorithm is simple and flexible. It is insensitive to data distribution and it is suitable for analog circuit fault feature extraction. The effectiveness of the algorithm is demonstrated by experimental results on analog circuit fault diagnosis examples.
出处
《电子测量与仪器学报》
CSCD
2007年第6期25-30,共6页
Journal of Electronic Measurement and Instrumentation
基金
铁道部科技计划项目(编号:2001X014)
关键词
特征提取
粒子群优化
K—NN分类
模拟电路故障诊断
feature extraction, particle swarm optimization, K-NN classification, analog circuit fault diagnosis.
