基于小波理论变换和神经网络的柴油机故障诊断方法的研究

提出一种基于小波理论和神经网络技术的柴油机振动诊断方法,首先对柴油机的振动信号进行小波分析,提取相应特征向量,然后将振动样本的特征向量作为RBF神经网络的输入参数,以故障类别作为输出参数训练该网络。训练后的神经网络可以利用测量的振动信号来判断柴油机的故障状况。试验及仿真证明该方法在柴油机振动诊断中是有效可行的,对其它复杂机械的振动诊断同样具有参考价值。

Vibration Diagnoses for Diesel Engine Based on Wavelet Analysis and Neural Networks

A new vibration diagnosis method for diesel engine based on wavelet theory and neural network is proposed. Firstly the wavelet theory is used to analyze vibration signal of diesel engine, mainly the wavelet decomposition and reduced noises are used to achieve the relevant characteristic vectors. Secondly model of neural networks is built, its input parameters include eight relevant characteristic vectors, and output parameters represent five kinds of faults. At last experiment data is used to train this RBF neural network. After that treatment, the RBFNN can be used to classify faults of diesel engine according to the input vibration signal exerted on diesel engine. The simulation and test results demonstrate that this method can efficiently diagnose and classify faults.