Task-optimal data-driven surrogate models for eNMPC via differentiable simulation and optimization
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929741924139008 |
|---|---|
| author | Mayfrank, Daniel Ahn, Na Young Mitsos, Alexander Dahmen, Manuel |
| author_facet | Mayfrank, Daniel Ahn, Na Young Mitsos, Alexander Dahmen, Manuel |
| contents | Mechanistic dynamic process models may be too computationally expensive to be usable as part of a real-time capable predictive controller. We 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 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 training algorithms 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 while eliminating 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_14425 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Task-optimal data-driven surrogate models for eNMPC via differentiable simulation and optimization Mayfrank, Daniel Ahn, Na Young Mitsos, Alexander Dahmen, Manuel Machine Learning Optimization and Control Mechanistic dynamic process models may be too computationally expensive to be usable as part of a real-time capable predictive controller. We 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 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 training algorithms 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 while eliminating 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. |
| title | Task-optimal data-driven surrogate models for eNMPC via differentiable simulation and optimization |
| topic | Machine Learning Optimization and Control |
| url | https://arxiv.org/abs/2403.14425 |