摘要
柴油机振动信号具有非平稳性,用最优小波包将不同故障的振动信号分解到不同频段。提取各频段的能量组成特征向量输入SOM-BP神经网络,通过神经网络输出结果判别柴油机的故障类型。与BP网络的训练结果相比较,证明将最优小波包分解与SOM-BP神经网络相结合的方法可以得到更好的分类结果,有一定的工程实用性。
Vibration signal of diesel engine is non-stationary,using optimal wavelet packet,different fault vibration signals were decomposed to different frequency bands.And the feature vectors composed of energy of different bands constitute input vectors of SOM-BP neural network.The types of engine failure were judged by output results of SOM-BP neural network.Compared to BP network,the method of combination of optimal wavelet packet and SOM-BP neural network is better,and experimental results confirmed practicality.
出处
《煤矿机械》
北大核心
2012年第10期278-280,共3页
Coal Mine Machinery
基金
国家自然科学基金资助项目(50875247)
教育部博士点基金资助项目(20091420110002)
山西省自然科学基金资助项目(2007011070)