Decreasing computational cost of simulation based interactive multiobjective optimization with adjustable solution accuracy
Julkaistu sarjassa
Jyväskylän yliopisto. Reports of the Department of Mathematical Information Technology. Series B. Scientific computingPäivämäärä
2008Solving real-life engineering problems can be time-consuming and difficult because problems may have multiple conflicting objectives, functions involved highly nonlinear and containing multiple local minima, and function values are often produced via a time-consuming simulation process. Problems of this type can be solved using global multiobjective optimization methods, preferably with interactive approaches, which allow the designer (or decision maker in general) to learn about the behaviour of the problem during the solution process. In an interactive approach the designer specifies preferences and Pareto optimal solution(s) following these preferences are generated, typically by forming a scalarizing function and solving it. In simulation based optimization this may take time. Thus, the designer may have to wait for a long before (s)he can continue the solution process. Although some efficient global optimization algorithms exist, it is of outmost importance to be able to reduce the computational burden. In our study, we show that substantial savings in calculation time can be achieved using a decreased number of function evaluations at the beginning of the interactive solution process, without compromising the quality of the final solution too much. Furthermore, at each iteration we use simple heuristics to judge sufficient amount for computation. As the designer has gained more understanding about the problem, (s)he may approach the final solution with an ever increasing accuracy and number of objective function evaluations. We show results using several different budget schemes for calculation, and identify levels where a sufficient quality for final solutions is retained.
...
ISBN
978-951-39-9035-0Metadata
Näytä kaikki kuvailutiedotKokoelmat
- Digitoidut julkaisut [264]
Samankaltainen aineisto
Näytetään aineistoja, joilla on samankaltainen nimeke tai asiasanat.
-
Approximation method for computationally expensive nonconvex multiobjective optimization problems
Haanpää, Tomi (University of Jyväskylä, 2012) -
Approximation through interpolation in nonconvex multiobjective optimization
Hartikainen, Markus (University of Jyväskylä, 2011) -
On solving computationally expensive multiobjective optimization problems with interactive methods
Ojalehto, Vesa (University of Jyväskylä, 2014) -
An interactive surrogate-based method for computationally expensive multiobjective optimisation
Tabatabaei, Mohammad; Hartikainen, Markus; Sindhya, Karthik; Hakanen, Jussi; Miettinen, Kaisa (Palgrave Macmillan Ltd., 2019)Many disciplines involve computationally expensive multiobjective optimisation problems. Surrogate-based methods are commonly used in the literature to alleviate the computational cost. In this paper, we develop an interactive ... -
E-NAUTILUS: A decision support system for complex multiobjective optimization problems based on the NAUTILUS method
Ruiz, Ana B.; Sindhya, Karthik; Miettinen, Kaisa; Ruiz, Francisco; Luque, Mariano (Elsevier BV * North-Holland; Association of European Operational Research Societies, 2015)Interactive multiobjective optimization methods cannot necessarily be easily used when (industrial) multiobjective optimization problems are involved. There are at least two important factors to be considered with any ...
Ellei toisin mainittu, julkisesti saatavilla olevia JYX-metatietoja (poislukien tiivistelmät) saa vapaasti uudelleenkäyttää CC0-lisenssillä.