通过分析车辆电源系统的信号特征,提出了基于小波包与隐马尔可夫相结合的故障诊断方法.利用小波包分解提取电源系统各种状态下的信号特征,基于模拟退火思想改进K均值算法选取HMM初值,用特征向量训练连续HMM,再用训练好的HMM进行电源系统的状态监测与故障诊断,实验结果表明用少量样本就能取得很好的诊断效果.%By analyzing the signal characteristics of vehicle power system, fault diagnosing method is proposed, which is combined with wavelet packet and bidden Markov model. With the features extracted from the various states of power system by wavelet packet decomposition, it selects the initial value of HMM by improved K means algorithm based on the simulated annealing, trains CHMM with the feature vector, which is used for condition monitoring and fault diagnosis of power system. The results show that it has accurate diagnosis through small training sample.
展开▼