基于多特征信息融合的截齿磨损程度识别研究

为实现截割过程中截齿磨损程度的在线监测和精准识别,提出一种基于多特征信息融合的采煤机截齿磨损程度识别方法。针对不同磨损程度截齿截割过程中的振动和声发射信号进行时域分析和小波包分析;针对两相邻磨损程度截齿的特征样本存在数据交集,增加系统识别难度的问题,构建最小模糊度优化模型并计算各特征参数的最优模糊隶属度函数,获取特征样本的最大隶属度。运用多特征数据样本对BP(back-propagation)神经网络识别模型进行学习和训练。实验结果表明,截齿磨损程度识别模型的判别结果和样本实际磨损类别相符,能够实现对截齿磨损程度的在线监测和精准识别。研究结果对于实际工程中截齿的监测和更换具有重要意义。

Research on wear degree recognition of picks based on multi feature information fusion

In order to realize the on line monitoring and accurate identification of picks wear degree in the cutting process, a new method based on multi feature information fusion was proposed for identifying shearer pick wear degree. The vibration and acoustic emission signals in the cutting process of different picks wear degree were analyzed by time domain analysis and wavelet packet analysis. According to the features of two adjacent pick wear degrees, there were data intersections for characteristic samples, which increased the difficulty of system identification. The optimal fuzzy membership function for each characteristic signal was calculated by using the least fuzzy optimization model, the maximum membership degree of the feature sample was obtained. The back propagation(BP) neural network recognition model was trained and learned by using multi feature data samples. The experimental results show that the results of network discrimination are consistent with actual wear level of test sample. It can accurately monitor and identify the type of picks wear. The research results have great significance to monitoring and replacement of picks in actual engineering.