To address the issue of over-reliance on manual inspections for diagnosing vehicle shaking states in traditional high-speed turnout areas,a novel diagnosis method based on generalized demodulation and Sparrow Search Algorithm optimized Support Vector Machine Model(SSA-SVM)is proposed.Firstly,the lateral acceleration of the vehicle body is decomposed using generalized demodulation,and the modal components of different frequencies are extracted.By integrating this component information with the track geometry information,the diagnostic characteristic indicators for vehicle shaking in the turnout area are further calculated.Secondly,the SSA-SVM model is used as the classification diagnosis model for vehicle shaking in the turnout area,and a corresponding diagnosis method is proposed.Finally,a case study using measured data from high-speed railway turnout area in China is conducted to validate the effectiveness of the method.The results show that compared with the diagnosis methods based on Back-Propagation algorithm model(BP),SVM model,Particle Swarm Optimization algorithm optimized Back-Propagation algorithm model(POS-BP)and Particle Swarm Optimization algorithm optimized Support Vector Machine model(POS-SVM),the proposed method achieves faster convergence speed,higher accuracy,and maintains a high diagnostic accuracy of 94.8％even with fewer features.
关键词
高速铁路/道岔区晃车/广义解调/时频分析/SSA-SVM
Key words
High-speed railway/Vehicle shaking in turnout area/Generalized demodulation/Time-frequency analysis/SSA-SVM
维普
万方数据