Database-Driven Iterative Learning for Building Temperature Control

doi:10.1109/TASE.2019.2899377

Database-Driven Iterative Learning for Building Temperature Control

2019-10-01

30

OA

AI

DOI：10.1109/TASE.2019.2899377

Original

Save

View PDF

Abstract

En 中文

Building interior temperatures are affected by the outdoor air temperature. Note that outdoor weather patterns are somewhat repetitive in nature and historical records of outdoor temperature are readily accessible; we present a data-driven iterative learning approach to improve the room temperature tracking performance over time. By comparing the short-term temperature forecast with the past data, chains of (nonconsecutive) days exhibiting similar outside temperature patterns can be identified. The corresponding building operation record (heat input and temperature output trajectories) may then be used in the iterative learning control (ILC) to update the input based on the past temperature tracking error. Multizone buildings are strictly passive from the heat input to temperature output in all zones. This property assures the convergence of the ILC iteration if the update gain is suitably bounded, without the need of an accurate model. This means that for each chain, the zone temperature deviation from the specified profile will converge to zero as the number of days in the chain grows (i.e., as more iterations of ILC are performed). Using a six-zone physical test bed with programmable ambient temperatures, we demonstrate the practicality of the proposed approach in multiple experimental trials. Additional longer-duration simulations are performed based on the actual temperature recorded in Orlando, FL, USA and New York, NY, USA over a two-year period. In all cases, ILC is shown to improve the tracking error in the presence of ambient temperature fluctuations. Note to Practitioners-This paper uses historic operating data and weather forecasting to improve the temperature tracking in multizone buildings. The proposed approach exploits the inherent passivity property of the building thermal model and, therefore, does not require the accurate model identification prior to implementation. As a result, this method may be conveniently retrofitted to any existing heating, ventilation, and air conditioning system, provided that the system is capable of adjusting the heat input (e.g., by controlling the supply air flow rate) and storing temperature sensor data. The algorithm uses weather prediction to match the upcoming ambient temperature variation with those experienced in the past. With the past operating data (control input and resulting temperature error) from the matched day, a feed-forward control is iteratively updated to improve the tracking performance. This paper includes the experimental results to demonstrate the efficacy of the approach in our test bed, and simulation examples to provide the insight into the real-world implementation. This approach may be extended to other external heat sources exhibiting repetitive patterns, e.g., occupancy, equipment, and solar input.

Keywords:

Buildings

Temperature distribution

Heating systems

Temperature sensors

Meteorology

Predictive models

Building automation systems

iterative learning control (ILC)

heating

ventilation

and air conditioning (HVAC) control

Database-Driven Iterative Learning for Building Temperature Control

Abstract

AI Summary

Journal

Organization

Cited Papers