作者: Jörg Bremer , Michael Sonnenschein
DOI: 10.1007/978-3-662-44440-5_14
关键词:
摘要: A comparably new application for support vector machines is their use meta-modeling the feasible region in constrained optimization problems. Applications have already been developed to problems from smart grid domain. Still, problem of a standardized integration such models into (evolutionary) algorithms was as yet unsolved. We present decoder approach that constructs mapping unit hyper cube learned model. Thus, are transferred unconstrained ones by space easier search. result artificial test cases well simulation results real power planning scenarios.