A new approach for online multiobjective optimization of mechatronic systems |
| |
Authors: | Katrin Witting Bernd Schulz Michael Dellnitz Joachim Böcker Norbert Fröhleke |
| |
Affiliation: | (1) Faculty of Computer Science, Electrical Engineering and Mathematics, University of Paderborn, 33095 Paderborn, Germany |
| |
Abstract: | We present a new concept for online multiobjective optimization and its application to the optimization of the operating point
assignment for a doubly-fed linear motor. This problem leads to a time-dependent multiobjective optimization problem. In contrast
to classical optimization where the aim is to find the (global) minimum of a single function, we want to simultaneously minimize
k objective functions. The solution to this problem is given by the set of optimal compromises, the so-called Pareto set. In
the case of the linear motor, there are two conflicting aims which both have to be maximized: the degree of efficiency and
the inverter utilization factor. The objective functions depend on velocity, force and power, which can be modeled as time-dependent
parameters. For a fixed point of time, the entire corresponding Pareto set can be computed by means of a recently developed
set-oriented numerical method. An online computation of the time-dependent Pareto sets is not possible, because the computation
itself is too complex. Therefore, we combine the computation of the Pareto set with numerical path following techniques. Under
certain smoothness assumptions the set of Pareto points can be characterized as the set of zeros of a certain function. Here,
path following allows to track the evolution of a given solution point through time. |
| |
Keywords: | Multiobjective online optimization Time-dependent Pareto sets Decision making Numerical path following methods Linear motor Operating point assignment Self-optimization |
本文献已被 SpringerLink 等数据库收录! |
|