Minimization of real-time factor in "miMA" vehicle-driver-road model

International Automotive Conference 2018

Maniowski, M.^1

Institute of Automobiles and Internal Combustion Engines, Cracow University of Technology, Poland^1

关键词: Computational time;    Full vehicle model;    Numerical accuracy;    Optimization algorithms;    Optimization task;    Ordinary differential equation (ODE);    Sampling frequencies;    Trial-and-error method;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/421/2/022022/pdf
DOI  :  10.1088/1757-899X/421/2/022022

来源: IOP

【 摘 要 】

The paper concerns further improvements at numerical level of computational time and accuracy of formulated by the Author model of vehicle-driver-road (called "miMA") for motorsport optimization applications. Selection of the best ordinary differential equations (ODE) solver (i.e.: ode23t, ode23s, ode15s, ode45) and its parameters (i.e.: sampling frequency, relative tolerance, absolute tolerance, maximum step) is carried out based on the formulated optimization algorithm, instead of trial and error methods. After optimization, it turned out, that the initial real-time factor (RTF = 0.20 for tsim = 1.6 s by using ode15s solver) of "miMA" model, can be reduced by even 50% (to RTF = 0.087 for tsim = 0.695 s by using ode23t solver ) without scarifying too much the model numerical accuracy. These are key features for effective solving of large optimization tasks, like a lap time minimization with full vehicle model.