基于小波包分解与决策树的滚子链状态检测研究

Research on Roller Chain State Detection Based on Wavelet Packet Decomposition and Decision Tree

本文提出一种基于小波包分解与决策树相结合的检测方法对滚子链的磨损状态进行检测。该方法首先使用小波包理论对信号进行n(n=3)层小波分解,提取小波系数,并根据系数求解各个频段的能量,并构造特征向量,然后使用决策树策略对滚子链的状态进行判断。实验表明,通过小波包分解和决策树相结合的检测方法能够很好地识别滚子链的磨损状态。

In this paper,a detection method based on wavelet packet decomposition and decision tree is proposed to detect the wear state of the roller chain. The method firstly uses wavelet packet theory to perform n(n=3) layer wavelet decomposition on the signal,extracts the wavelet coefficients,and solves the energy and structural feature vectors of each frequency band according to the coefficients,and then uses the decision tree strategy to judge the state of the roller chain. Experiments show that the detection method combined with wavelet packet decomposition and decision tree can well identify the wear state of the roller chain.