基于PSO-SVM的板式无砟轨道CA砂浆脱空损伤识别

Contact loss identification of CA mortar in prefabricated slab track based on PSO-SVM

为准确识别板式无砟轨道层间脱空损伤,利用有限元软件ANSYS/LS-DYNA建立含砂浆损伤的车辆-板式无砟轨道耦合动力学模型,提取基于振动响应的损伤特征指标,并利用主成分分析法对其进行降维处理,将获得的脱空样本输入基于粒子群优化算法-支持向量机(PSO-SVM)搭建的脱空识别模型进行损伤识别。研究结果表明:从时域、频域及时-频域角度对振动信号提取的26个特征指标,可较全面地反映振动信号中隐含的损伤信息;利用主成分分析法可在尽可能保留原始损伤信息的基础上,有效降低损伤特征指标的维度,提高损伤识别效率;采用本文提出的PSO-SVM识别算法对砂浆脱空识别是可靠、有效的,识别准确率达到90%。

To identify the interlayer contact loss of the prefabricated slab track accurately, a vehicle-slab track coupling dynamic model considering CA mortar damage was established by using ANSYS/LS-DYNA finite element software. Damage characteristic indexes were extracted based on vibration response and its dimension was reduced using principal component analysis. And the obtained contact loss samples were input into the identification model based on the particle swarm optimization algorithm-support vector machine(PSO-SVM) for damage identification. The results show that the 26 characteristic indexes extracted from the vibration signal in the time domain, frequency domain and time-frequency domain can fully reflect the hidden damage information. The principal component analysis method can effectively reduce the characteristic index dimensions and improve damage identification efficiency. The PSO-SVM identification algorithm proposed in this paper is reliable and efficient for mortar contact loss identification, and the identification accuracy reaches 90%.