An embedded industrial control framework for model predictive control of district heat substation

Abstract

In this paper we present a standard platform XC05 for an Edge Controller based on an Industrial Control System, where functions made in Modelica and Python can be run as an integrated part of an automation system. We demonstrate how the platform is used to run a complex Model Predictive Control (MPC) strategy to optimize indoor heating in a residential building. MPC strategies have been increasingly popular due to their ability to handle nonlinear dynamics with constraints and multi-objective optimization. Since industrial control systems are real-time based, consideration must also be taken to running security and the real-time characteristics and timing of the overall system solution. We also show that heavy calculation, protected by the industrial control system operative, can run safely together within fast automation using standard electronics. The controlled variable in the MPC strategy is the supply water temperature (Space heating), and the objective is to keep the indoor temperature at a predefined setpoint despite variations in outdoor weather conditions by using local measurements and weather forecasts from the Swedish weather service SMHI. The model used in the MPC is trained automatically with real-time data during running. We describe the controller architecture and briefly the model predictive control algorithm, analyze the overall system performance regarding safety and real-time characteristics. The proposed model predictive control application showed stable operation and expected real-time characteristics during operation. Furthermore, a reduction in indoor temperature deviations was achieved.