001038126 001__ 1038126
001038126 005__ 20250203103312.0
001038126 0247_ $$2doi$$a10.48550/ARXIV.2403.14425
001038126 037__ $$aFZJ-2025-01173
001038126 1001_ $$0P:(DE-Juel1)192151$$aMayfrank, Daniel$$b0$$ufzj
001038126 245__ $$aTask-optimal data-driven surrogate models for eNMPC via differentiable simulation and optimization
001038126 260__ $$barXiv$$c2024
001038126 3367_ $$0PUB:(DE-HGF)25$$2PUB:(DE-HGF)$$aPreprint$$bpreprint$$mpreprint$$s1738063419_6967
001038126 3367_ $$2ORCID$$aWORKING_PAPER
001038126 3367_ $$028$$2EndNote$$aElectronic Article
001038126 3367_ $$2DRIVER$$apreprint
001038126 3367_ $$2BibTeX$$aARTICLE
001038126 3367_ $$2DataCite$$aOutput Types/Working Paper
001038126 520__ $$aWe present a method for end-to-end learning of Koopman surrogate models for optimal performance in a specific control task. In contrast to previous contributions that employ standard reinforcement learning (RL) algorithms, we use a training algorithm that exploits the potential differentiability of environments based on mechanistic simulation models to aid the policy optimization. We evaluate the performance of our method by comparing it to that of other controller type and training algorithm combinations on an existing economic nonlinear model predictive control (eNMPC) case study of a continuous stirred-tank reactor (CSTR) model. Compared to the benchmark methods, our method produces similar economic performance but causes considerably fewer and less severe constraint violations. Thus, for this case study, our method outperforms the others and offers a promising path toward more performant controllers that employ dynamic surrogate models.
001038126 536__ $$0G:(DE-HGF)POF4-1121$$a1121 - Digitalization and Systems Technology for Flexibility Solutions (POF4-112)$$cPOF4-112$$fPOF IV$$x0
001038126 536__ $$0G:(DE-Juel1)HDS-LEE-20190612$$aHDS LEE - Helmholtz School for Data Science in Life, Earth and Energy (HDS LEE) (HDS-LEE-20190612)$$cHDS-LEE-20190612$$x1
001038126 588__ $$aDataset connected to DataCite
001038126 650_7 $$2Other$$aMachine Learning (cs.LG)
001038126 650_7 $$2Other$$aOptimization and Control (math.OC)
001038126 650_7 $$2Other$$aFOS: Computer and information sciences
001038126 650_7 $$2Other$$aFOS: Mathematics
001038126 7001_ $$0P:(DE-HGF)0$$aAhn, Na Young$$b1
001038126 7001_ $$0P:(DE-Juel1)172025$$aMitsos, Alexander$$b2$$ufzj
001038126 7001_ $$0P:(DE-Juel1)172097$$aDahmen, Manuel$$b3$$eCorresponding author$$ufzj
001038126 773__ $$a10.48550/ARXIV.2403.14425
001038126 909CO $$ooai:juser.fz-juelich.de:1038126$$pVDB
001038126 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)192151$$aForschungszentrum Jülich$$b0$$kFZJ
001038126 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-HGF)0$$aForschungszentrum Jülich$$b1$$kFZJ
001038126 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)172025$$aForschungszentrum Jülich$$b2$$kFZJ
001038126 9101_ $$0I:(DE-588b)36225-6$$6P:(DE-Juel1)172025$$aRWTH Aachen$$b2$$kRWTH
001038126 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)172097$$aForschungszentrum Jülich$$b3$$kFZJ
001038126 9131_ $$0G:(DE-HGF)POF4-112$$1G:(DE-HGF)POF4-110$$2G:(DE-HGF)POF4-100$$3G:(DE-HGF)POF4$$4G:(DE-HGF)POF$$9G:(DE-HGF)POF4-1121$$aDE-HGF$$bForschungsbereich Energie$$lEnergiesystemdesign (ESD)$$vDigitalisierung und Systemtechnik$$x0
001038126 9141_ $$y2024
001038126 920__ $$lyes
001038126 9201_ $$0I:(DE-Juel1)ICE-1-20170217$$kICE-1$$lModellierung von Energiesystemen$$x0
001038126 980__ $$apreprint
001038126 980__ $$aVDB
001038126 980__ $$aI:(DE-Juel1)ICE-1-20170217
001038126 980__ $$aUNRESTRICTED