In this paper, we compared several kinds of matrix libraries on a real-time analysis environment. The computational speed of multibody dynamic analysis depends on the number of bodies and constraints and the eigenfrequency of the system to be analyzed, and it is necessary to enhance the computational efficiency to realize real-time analysis of a large size system or of high frequency phenomena. Our real-time simulation environment of this study was with xPC Target, which is one of the MATLAB products. First, we evaluated the performance of matrix libraries by solving large-size linear equation consisting of random numbers. Next, the matrix libraries were embedded to real-time multibody dynamics code developed by authors and the computational time was evaluated. The simulation model was multiple pendulums model and we investigated relationship between calculation performance and the number of bodies. Finally we analyzed vehicle dynamics as an example of actual multibody system, and the performance of using matrix libraries is discussed.
[1]
E. J. Haug,et al.
Computer aided kinematics and dynamics of mechanical systems. Vol. 1: basic methods
,
1989
.
[2]
Jack Dongarra,et al.
Numerical Linear Algebra for High-Performance Computers
,
1998
.
[3]
Yoshihiro Suda,et al.
Evaluation of driver’s behavior with multibody-based driving simulator
,
2007
.
[4]
Taichi Shiiba,et al.
Experimental validation of steering torque feedback simulator through vehicle running test
,
2009
.
[5]
Daniel Dopico,et al.
PARALLEL LINEAR EQUATION SOLVERS AND OPENMP IN THE CONTEXT OF MULTIBODY SYSTEM DYNAMICS
,
2009
.