COSTARICA estimator for rollback-less systems handling in iterative co-simulation algorithms
Co-simulation is widely used in the industry due to the emergence of modular dynamical models made up of interconnected, black-boxed systems. Several co-simulation algorithms have been developed, each with different properties and different levels of accuracy and robustness. Among them, the most accurate and reliable ones are the iterative ones, although they have a main drawback in common: the involved systems are required to be capable of rollback. The latter denotes the ability of a system to integrate over a co-simulation time step that has already been simulated. Non-rollback-capable system can only go forward in time and every integrated step is definitive. In practice, the industrial modelling and simulation platforms rarely produce rollback-capable systems. This paper proposes a solution that slightly changes the co-simulation methodology and that enables to use iterative co-simulation methods on a modular model which contains non-rollback-capable systems in case the latter represent ordinary differential equations. The idea is to replace such a system by a simplified version, which is used to estimate the results of the integrations instead of integrating the real system. Once the co-simulation method's surrogate iterations on these estimators predict the convergence on the co-simulation step, the non-rollback-capable systems genuinely integrate the step using the estimated solution on the other systems before moving forward, transforming the iterative co-simulation method into a non-iterative one.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Generator-Assistant Stepwise Rollback Framework for Large Language Model Agent
Large language model (LLM) agents typically adopt a step-by-step reasoning framework, in which they interleave the processes of thinking and acting to accomplish the given task. However, this paradigm faces a deep-rooted…
Decision MakingLanguage ModelingLanguage ModellingLarge Language ModelLearning to Undo: Rollback-Augmented Reinforcement Learning with Reversibility Signals
This paper proposes a reversible learning framework to improve the robustness and efficiency of value based Reinforcement Learning agents, addressing vulnerability to value overestimation and instability in partially irr…
Reinforcement LearningDecision MakingGoverned Capability Evolution: Lifecycle-Time Compatibility Checking and Rollback for AI-Component-Based Systems, with Embodied Agents as Case Study
Software systems built from versioned AI components increasingly need lifecycle-time governance: when a capability module evolves into a new version, the hosting system must decide whether the new version may be activate…
ChronoMem: Version Control and Semantic Rollback for Large Language Model Agent Memory
LLM agents increasingly rely on long-term memory to support multi-session interaction and personalization. However, existing agent memory systems are designed around forward-only evolution, continuously accumulating, con…
Semantic RetrievalQuestion AnsweringRENet: Fault-Tolerant Motion Control for Quadruped Robots via Redundant Estimator Networks under Visual Collapse
Vision-based locomotion in outdoor environments presents significant challenges for quadruped robots. Accurate environmental prediction and effective handling of depth sensor noise during real-world deployment remain dif…