论文部分内容阅读
将分形理论引入车用发动机运行状态评估及趋势预测中,旨在利用分形的特征参数-关联维数来定量描述、解释不同状态的振动信号,探讨关联维数与发动机不同运行状态的内在关系。本文对多组不同状态的振动信号进行相空间重构,计算出对应的关联维数,结果表明,关联维数能定量评估发动机的状态演化,敏感反映运行状态的异常变化。为了实现及时维修,本文同时利用改进的BP神经网络对邻近状态进行预测,取得很好的效果。
The fractal theory is introduced into the running state evaluation and trend prediction of vehicle engine. The purpose is to use the fractal characteristic parameter - correlation dimension to quantitatively describe and explain the vibration signals of different states and to explore the relationship between the correlation dimension and different operating states of the engine. In this paper, the phase space of several vibration signals of different states is reconstructed, and the corresponding correlation dimension is calculated. The results show that the correlation dimension can quantitatively evaluate the evolution of the engine and reflect the abnormal changes of operating conditions sensitively. In order to achieve timely maintenance, this paper also uses the improved BP neural network to predict the neighboring states and achieves good results.