A reliable mixed-integer linear programming formulation for data-driven model predictive control in buildings

TL;DR

This paper introduces a new control strategy for buildings with thermal structures and shading systems, using a reliable optimization method to improve energy efficiency and comfort.

Contribution

A novel MILP-based DMPC formulation for buildings with TABS and shading systems using a grey-box thermal model.

Findings

The proposed DMPC formulation handles thermal comfort constraints effectively.

The grey-box model reduces the need for detailed white-box modeling.

The approach improves replicability and efficiency in building energy control.

Abstract

Integrating renewable energy sources into buildings requires advanced control strategies to enhance demand-side flexibility. Data-driven Model Predictive Control (DMPC) has shown significant promise in this area. Buildings with Thermally Activated Building Structures (TABS) and glass façade present flexibility potential, but have a challenging thermal balance, due to high thermal inertia and significant solar gains. In such buildings, a DMPC jointly controlling TABS and the shading system is advantageous. However, the only known implementations enabling this combination rely on a white-box model, limiting the replicability of the approach due to the required modelling effort. To facilitate the implementation of a DMPC for such a combined control task, this paper presents in details a suitable optimisation algorithm:•specifically designed for buildings with TABS and shading systems,•based on a grey-box model of thermal zones (a reduced order state-space model),•using a reliable and efficient Mixed-Integer Linear Programming (MILP) formulation,•handling thermal comfort as constraints, avoiding weighting factors in the objective function.

specifically designed for buildings with TABS and shading systems,

based on a grey-box model of thermal zones (a reduced order state-space model),

using a reliable and efficient Mixed-Integer Linear Programming (MILP) formulation,

handling thermal comfort as constraints, avoiding weighting factors in the objective function.

Image, graphical abstract

Figures