Accompanying code repository to our work "Probabilistic multi-step identification with implicit state estimation for stochastic MPC"

Introduction

Dear reader,

welcome to this repository. You'll find here the code that was created to produce the results for our work "Probabilistic multi-step identification with implicit state estimation for stochastic MPC". Please don't hesitate to write us a message, if you have any questions. If your questions is of concern to other users, we suggest you use the discussions tab. To better understand the structure of this repository and our code please read the overview below.

Structure

This repository is structured as follows.

Under blrsmpc you'll find the entire source code that was used to create the results. In particular:

• system: The ODE and discretizte formulation of the investigated building system model and our simulation tools
• sysid: The required code for probabilistic identification of state-space and multi-step models
• SMPC: Our implementation of an SMPC controller with state-space and multi-step model

Our results can be recreated with the code in results. We show two examples:

• A linear building system
  • In sid_building_compare_kf_pred, the code to create Figure 1 can be found.
  • In sid_building_output_fb_for_smpc, we identify the multi-step and state-space model for SMPC
  • In smpc_meta_building, we perform SMPC with the identified models and the results in Figure 2 and Table 1 are created.
• A nonlinear CSTR system
  • In sid_cstr, we identify the multi-step and state-space model for the SMPC. The results shown in Figure 3 are created.
  • In smpc_cstr, the results shown in Figure 4 and Table 3 are created.

We strive to make our results as accesible as possible and are happy to get feedback.