【作者】 于开平； 李静； 庞世伟； 邹经湘；

【Author】 YU Kai-ping, LI Jing, PANG Shi-wei, ZOU Jing-xiang( Dept. of Astronautic Eng. and Mechanics, Ha’erbin University of Industry, Ha’erbin 150001; Aeronautic Technique and Engineering Group of China, Beijing 100088)

【机构】 哈尔滨工业大学航天工程与力学系； 中国航天科工集团二院二部；

【摘要】 本文系统地研究了基于遗传算法的线性结构系统模态参数识别方法。遗传算法对广义动力学系统的辨识已经得到了充分研究,但在结构系统尤其是时变结构系统参数辨识方面的研究还很少,前期的工作主要集中于二进制编码的遗传算法的改进。对于数值优化问题,使用浮点数编码更接近问题的“自然表示”。因此,本文使用浮点数编码方案,分别利用频响函数和脉冲响应函数数据,给出了频域和时域方法两种辨识方法。由于遗传算法的全局寻优特性,该算法具有极强的抗噪声能力,即使对噪声严重污染的数据依然能够很好的辨识出待辨识的参数。通过一个两自由度的弹簧质量块系统对线性时不变情况分别进行了仿真分析,分析结果验证了以上结论的正确性。更多还原

【Abstract】 This paper studied the linear structure system modal parameters identification based on Genetic Algorithms. Genetic Algorithms has been well studied in generalization dynamics system identification, but less work has been done in the structural system identification. The previous study focused on the improvement of the Genetic Algorithms based on binary coding. In fact, floating point coding is the "best nature" description of numerical optimization. Therefore, based on floating point coding, the author obtained two kinds of different new methods -frequency domain method and time domain method, through using the different data- frequency response data and impulse response data. Because of the global optimization of Genetic Algorithms, both method has good quality of resisting noise. By simulating of a two degree-of-freedom mass-spring-system, author studys the application of these methods in linear time-invarying system. The simulation results testify the conclusion.更多还原