中文摘要：

针对LuGre摩擦模型参数耦合、动态参数辨识困难等问题，提出一种基于区间分析的LuGre摩擦模型动态参数辨识方法。首先，简要介绍了边界误差估计方法及其在系统模型为常微分方程时系统参数辨识中的应用，然后，基于两步法思想，分别采用最小二乘法和基于区间分析的边界误差估计方法对4个静态参数和2个动态参数进行了辨识。该方法有效避免了采用传统非线性辨识方法辨识LuGre摩擦模型动态参数时初值确定困难、对辨识结果影响大以及显性目标函数不易给出等缺陷，同时，辨识得到的参数为全局最优。最后通过实例验证了该方法的正确性和有效性。

英文摘要：

An approach, based on bounded-error estimation via interval analysis, was presented to identify dynamic parameters of LuGre friction model herein. The approach assumed the error of measurement output data of dynamic system with LuGre friction force was bounded. Hence, the problem of parameter idelification can be transformed into that of set inversion, which will be effec- tively solved via interval analysis. The four static parameters of LuGre friction model were estimated using least--square method in sliding regime, which deceased sharply the amounts of parameters iden- tified using interval analysis. In comparison with the traditional estimation methods the main advanta- ges of the proposed method are that, as it is global, it bypasses the problem of initialization. Moreo-ver, the method avoids amount error due to transformation of measurement output data. An identifi-cation example was given to show the correctness and efficiency of the proposed method.