基于多视图分类集成的高铁工况识别
Condition recognition of high-speed train based on multi-view classification ensemble

摘要： 针对传统方法识别高铁工况存在特征提取不完备和识别性能不精确的问题,提出一种多视图分类集成的高铁工况识别方法(MVCE)。该方法结合多视图特征提取和分类集成技术,从信号本身特性、频域和时频域三个角度提取小波能量、频谱系数、聚合经验模态分解模糊熵,并使用Fisher比率对其频域特征进行特征选择,从而构建高铁振动信号三个视图的特征。使用最小二乘支持向量机和K最近邻分类器分别对每个视图的特征进行初步识别。最后采用分类熵投票策略对多个分类器输出结果进行集成。试验结果表明:该方法对仿真数据和实验室数据的平均识别率分别达到89.18%和90.87%。同时对比结果说明了该方法提取特征的完备性和具有多样性集成模型的有效性。
Abstract: To solve the problem about the incompletion of feature extraction and inaccuracy of the identification performance of traditional method, a multi-view classification ensemble method(MVCE)for condition recognition of high speed train was proposed. The method combined with multi-view feature extraction and classification ensemble technology. For condition recognition, the wavelet energy, spectral coefficients and ensemble empirical mode decomposition fuzzy entropy were extracted from three angles: the characteristics of the signal, the frequency domain and the time-frequency domain. The Fisher ratio was used to perform feature selection for the frequency domain features of the high speed train vibration signal, then the feature of the three views were constructed collectively. The least square support vector machine(LSSVM)and the K nearest neighbor(KNN)classifiers were used to recognize each view. The output results of multiple classifiers were integrated by using the classification entropy voting principle. The experimental results showed that the average recognition rate of the proposed method on the simulation data and the laboratory data were 89.18% and 90.87% respectively. Meanwhile, the comparative results illustrated the completeness of the features extracted by the method and the validity of the ensemble model with diversity