摘要
结合HHT同GA-BP神经网络的优点,提出了将二者结合用于风机故障诊断的新方法,并且应用最新提出的方法改进HHT(希尔伯特-黄变换),使其应用更为有效。利用HHT构造出代表振动信号特征的"能-频分布";根据GA-BP网络模型能够逼近任意非线性函数和具有高效寻找全局最优的特点作为特征分类器,进行故障诊断。风机故障诊断结果表明,该方法是可行有效的。
Based on the advantage of the Hilbert-Huang Transform（HHT） and GA-BP neural network（BP neural network improved by Genetic Algorithms）,a new method of fault diagnosis for centrifugal fan combined the two methods was put forward and some methods put forward recently for the improvement of HHT were also applied here to make HHT more efficient.The ＂Energy-Frequency Distribution＂ of vibration signals was built by HHT,served as feature vectors of centrifugal fan vibration signals.A GA-BP model of neural network（GA-BPNN） acted as a classifier for fault diagnosis based on the fact that it can well approach any nonlinear continuous function and high efficiency in finding the optimization approach reflect dynamic features of the systems. The experimental result indicates that this method was efficient and feasible.
出处
《风机技术》
2010年第5期48-53,共6页
Chinese Journal of Turbomachinery
