Virtual Surface Temperature Sensor for Multi-zone Commercial Buildings

Multi-zone structure is commonly used in small commercial office buildings, retail stores and supermarket. While there is no adjacent partition wall between the zones, the impact of a neighbor zone on the current zone can be approximated and analyzed through the application of virtual walls. It is critical to accurately estimate the virtual wall surface temperature in order to evaluate the model uncertainty and apply improved supervisory control on multiple rooftop air-conditioning units (RTUs). We propose an innovative virtual surface temperature sensor based on system-identification to solve this challenge. The validation of the virtual temperature model is processed by the three validation criteria: goodness of fit (G), mean squared error (MSE) and coefficient of determination (R2) through off-control conditions with data obtained from a building simulation platform. Further, the sensitivity analysis using the on-control three conditions (under-sizing, properly sizing and oversizing condition) is conducted for analyzing and evaluating the performance of this system-identification based virtual sensor.