摘要
本文提出了基于果蝇算法优化支持向量机的分类方法。该方法利用风机振动频域的特征向量作为学习样本,然后运用改进的支持向量机模型对风机振动信号的故障特征进行模式识别。同时还运用了蚁群和粒子群两种智能算法,对向量机进行了优化,仿真结果表明,基于果蝇优化的最小二乘支持向量机方法具有识别率高,诊断速度快的优点,该方法是可行有效的。
This paper presents the classification based on drosophila algorithm optimizing SVM. The method utilizes feature vectors of frequency domain of fan vibration as learning samples. Then the improved support vector machine model is used to recognize the pattern of fault characteristic of fan vibration signal. This paper also uses the ant colony and particle swarm two kinds of intelligent algorithms to optimize support vector machines. Simulation results show that the method of drosophila optimization least squares support vector machine has a high recognition rate and fast diagnose speed. This method is feasible and effective
出处
《风机技术》
2014年第6期50-55,共6页
Chinese Journal of Turbomachinery
关键词
离心式风机
故障诊断
最小二乘支持向量机
果蝇算法
centrifugal fan
fault diagnose
least squares support vector machine
drosophila optimization
