基于非负矩阵分解的高速列车走行部工况识别

根据高速列车振动信号数据的特点,针对高速列车走行部工况的识别,提出了一种基于非负矩阵分解(NMF)算法的工况识别模型。该模型通过数据处理获得具有辨识度的工况数据集,使用K-meams、谱聚类、近邻传播和模糊C均值4种聚类算法获得初步识别结果。聚类算法的初步识别结果作为NMF算法的输入,获得最终识别结果。相比NMF算法与聚类算法的识别结果,NMF算法具有更高的准确率,本文验证了NMF算法在高速列车工况识别方面的有效性和高效性,为高速列车的工况识别提出了一种有效的解决方案。

Working Condition Recognition of Vehicle Bogies of high-speed Trains Based on Non-negative Matrix Factorization Algorithm

According to the data characteristics of vibration signal of high speed train,a working condition recognition model based on non negative matrix factorization algorithm is proposed for recognizing the working condition of vehicle bogie of high-speed train.A recognizable data set can obtain after data processing by the model.The preliminary recognition results can be obtained by using the four clustering algorithms of K means,spectral clustering,neighbor propagation and fuzzy C-means.And.the preliminary recognition result of the clustering algorithm is further used as the input of the NMF algorithm to obtain the final recognition result.Compared with the rec ognition results of NMF algorithm and clustering algorithm,the NMF algorithm has higher accu­racy.Hence,the effectiveness of the non-negative matrix algorithm for recognizing the working condition of high speed train is verified,providing an effective solution for recognizing the work ing condition of high speed trains.