# Safe model-based design of experiments using Gaussian processes

arXiv: https://arxiv.org/abs/2011.10009 · HTML: https://mlsanity.com/paper/2011.10009/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 19 Nov 2020.

## Built on by (most cited first, 2 of 2)

- [Safe Chance Constrained Reinforcement Learning for Batch Process Control](https://mlsanity.com/paper/2104.11706/) (2021)
- [Multi-fidelity Gaussian Process for Biomanufacturing Process Modeling with Small Data](https://mlsanity.com/paper/2211.14493/) (2022)

## Abstract (arXiv)

Construction of kinetic models has become an indispensable step in the development and scale up of processes in the industry. Model-based design of experiments (MBDoE) has been widely used for the purpose of improving parameter precision in nonlinear dynamic systems. This process needs to account for both parametric and structural uncertainty, as the feasibility constraints imposed on the system may well turn out to be violated leading to unsafe experimental conditions when an optimally designed experiment is performed. In this work, a Gaussian process is utilized in a two-fold manner: 1) to quantify the uncertainty realization of the physical system and calculate the plant-model mismatch, 2) to compute the optimal experimental design while accounting for the parametric uncertainty. This method provides a guarantee for the probabilistic satisfaction of the constraints in the context of model-based design of experiments. The method is assisted with the use of adaptive trust-regions in order to facilitate a satisfactory local approximation. The proposed method is able to allow the design of optimal experiments starting from limited preliminary knowledge of the parameter set, leading to a safe exploration of the parameter space. The performance of this method is demonstrated through illustrative case studies regarding the parameter identification of the kinetic model in flow reactors.