Purpose

The excitation force of under-chassis active equipment of railway vehicles has a significant impact on the flexible vibration of the car body. Mastering the excitation force of equipment can control the vehicle body vibration more comprehensively.

Methods

To identify the excitation force of under-chassis equipment of railway vehicles, four frequency-based force identification methods such as least square method and Tikhonov regularization methods based on L-Curve, ordinary cross-validation, and generalized cross-validation respectively, are studied.

Results and Conclusions

Laboratory test shows the effectiveness of the above identification methods. A plate model is used to analyze the influence of the measuring positions and measuring noise on the identification accuracy of different types of excitation forces. It is found that it is a suitable method to determine the optimal combination of measuring points according to the optimal condition number criterion. Based on the finite element model of the car body with under-chassis equipment, the excitation force of the under-chassis equipment is identified, and the validity of the optimal condition number criterion is verified. In addition, the results show that when the frequency response function and the responses of measuring points are disturbed by noise, the above methods show different robustness to the models with different degrees of complexity. The equipment structure of railway vehicles is complicated, which is easy to be disturbed by noise in the actual test, and the signal-to-noise ratio is small. Therefore, it may be appropriate to choose Tikhonov regularization based on the L-Curve.

中文翻译：

---

铁路车辆底盘设备激励力的频域识别

目的

铁路车辆底盘主动设备的激振力对车身的柔性振动有重大影响。掌握设备的激振力可以更全面地控制车身振动。

方法

为了识别铁路车辆底盘设备的激励力，分别研究了基于L曲线的最小二乘法和Tikhonov正则化方法，普通交叉验证和广义交叉验证这四种基于频率的力识别方法。 。

结果与结论

实验室测试证明了上述鉴定方法的有效性。平板模型用于分析测量位置和测量噪声对不同类型的激励力的识别精度的影响。发现一种根据最优条件数准则确定测量点最优组合的合适方法。基于具有底架设备的车身的有限元模型，确定了底架设备的激振力，并验证了最优条件数准则的有效性。此外，结果表明，当频率响应函数和测量点的响应受到噪声干扰时，上述方法对于具有不同复杂度的模型表现出不同的鲁棒性。铁路车辆的设备结构复杂，实际测试中容易受到噪声的干扰，信噪比小。因此，基于L曲线选择Tikhonov正则化可能是合适的。