02381nam 2200385z- 450 991080568060332120260217113301.0(CKB)5720000000263540(oapen)doab82583(IL-JeEL)995720000000263540(EXLCZ)99572000000026354020240213c2022uuuu -u- |gerurmn|---annantxtrdacontentcrdamediacrrdacarrierModellfreies Lernen optimaler zeitdiskreter Regelungsstrategien für Fertigungsprozesse mit endlichem Zeithorizont /Johannes DornheimKIT Scientific Publishing20221 online resource (226 p.)Schriftenreihe des Instituts für Angewandte Materialien, Karlsruher Institut für Technologie3-7315-1158-4 9783731511588 Includes bibliographical references and index.The quality and performance of components is largely determined by the execution of the manufacturing processes involved. The process result depends -- in addition to the initial state of the component and the process -- on the course of the process. In many manufacturing processes, the course of the process can be decisively determined by manipulated variables that change over time. This work deals with methods to optimize these time-varying quantities under fluctuating process conditions. The quality of components depend to a large extent on the execution of the industrial processes involved in manufacturing. In addition to the initial conditions of the component and the process, the process result depends on the course of the process, which often can be significantly determined by time-varying manipulated variables. Methods for the optimization of these time-dependent quantities with regard to the component quality and depending on process conditions is the subject of this work.Reinforcement learningReinforcement learningCongressesReinforcement learning.Reinforcement learning671.34Dornheim Johannes1346523BOOK9910805680603321Modellfreies Lernen optimaler zeitdiskreter Regelungsstrategien für Fertigungsprozesse mit endlichem Zeithorizont3075627UNINA