Affiliations: [a] State Key Laboratory for Strength and Vibration of Mechanical Structures, Xi'an Jiaotong University, Xi'an, Shaanxi, China | [b] College of Computer Science and Technology, HuaQiao University, Xiamen, Fujian, China | [c] College of Mechanical and Automation, HuaQiao University, Xiamen, Fujian, China
Correspondence:
[*]
Corresponding author: Cheng Wang, State Key Laboratory for Strength and Vibration of Mechanical Structures, Xi'an Jiaotong University, No. 28, Xianning West Road, Xi'an 710049, Shaanxi, China. E-mail:cheng.wang@stu.xjtu.edu.cn
Abstract: Due to the ill-conditioned inverse characteristics of
uncorrelated multi-source random dynamic load identification problem, there
are large condition number and large identification errors for classic
least-squares of generalization method at inherent natural frequencies. In
order to avoid its illness and singularity, this multi-objective
optimization inverse problem is turned into single-objective optimization
forward problem by criterion function of minimization maximum relative
errors of all response measuring points, and we adopt genetic algorithm to
search this optimal solution then. Results of uncorrelated multi-source
vibration load identification on cylindrical shell CAE simulation data set
show that this new method is much better in precision and is less sensitive
for measurement noise than classic least-squares of generalization method.
Keywords: Random dynamic load identification, uncorrelated multi-source, frequency domain, multi-objective optimization, matrix least-squares inverse, minimization maximum relative errors, genetic algorithm
