arXiv subjects

Arne Speerforck

Publications and source records attributed to Arne Speerforck.

3 recordsLinked to original sources

arXiv2026

Model Predictive Control of Thermo-Hydraulic Systems Using Primal Decomposition

Decarbonizing the global energy supply requires more efficient heating and cooling systems. Model predictive control enhances the operation of cooling and heating systems but depends on accurate system models, often based on control volumes. We present an automated framework including time discretization to generate model predictive controllers for such models. To ensure scalability, a primal decomposition exploiting the model structure is applied. The approach is validated on an underground heating system with varying numbers of states, demonstrating the primal decomposition's advantage regarding scalability.

eess.SY↗

arXiv2025

Balanced Neural ODEs: nonlinear model order reduction and Koopman operator approximations

Variational Autoencoders (VAEs) are a powerful framework for learning latent representations of reduced dimensionality, while Neural ODEs excel in learning transient system dynamics. This work combines the strengths of both to generate fast surrogate models with adjustable complexity reacting on time-varying inputs signals. By leveraging the VAE's dimensionality reduction using a nonhierarchical prior, our method adaptively assigns stochastic noise, naturally complementing known NeuralODE training enhancements and enabling probabilistic time series modeling. We show that standard Latent ODEs struggle with dimensionality reduction in systems with time-varying inputs. Our approach mitigates this by continuously propagating variational parameters through time, establishing fixed information channels in latent space. This results in a flexible and robust method that can learn different system complexities, e.g. deep neural networks or linear matrices. Hereby, it enables efficient approximation of the Koopman operator without the need for predefining its dimensionality. As our method balances dimensionality reduction and reconstruction accuracy, we call it Balanced Neural ODE (B-NODE). We demonstrate the effectiveness of this methods on several academic and real-world test cases, e.g. a power plant or MuJoCo data.

cs.LG↗

arXiv2023

Integrated Planning of Multi-energy Grids: Concepts and Challenges

In order to meet ever-stricter climate targets and achieve the eventual decarbonization of the energy supply of German industrial metropolises, the focus is on gradually phasing out nuclear power, then coal and gas combined with the increased use of renewable energy sources and employing hydrogen as a clean energy carrier. While complete electrification of the energy supply of households and the transportation sector may be the ultimate goal, a transitional phase is necessary as such massive as well as rapid expansion of the electrical distribution grid is infeasible. Additionally, German industries have expressed their plans to use hydrogen as their primary strategy in meeting carbon targets. This poses challenges to the existing electrical, gas, and heating distribution grids. It becomes necessary to integrate the planning and developing procedures for these grids to maximize efficiencies and guarantee security of supply during the transition. The aim of this paper is thus to highlight those challenges and present novel concepts for the integrated planning of the three grids as one multi-energy grid.

eess.SY↗