Architectural Engineering



Virtual Partition Surface Temperature Sensor based on Linear Parametric Model

Date of this Version



D. Woradechjumroen, Y. Yu*, and H. Li. (2015) "Virtual Partition Surface Temperature Sensor based on Linear Parametric Model". Applied Energy. 2015, Vol.162: 1323-1335.


Copyright 2015 Elsevier Ltd.

Link goes to ScienceDirect site.


Multi-zone structure is common in commercial office buildings, retail stores and supermarkets. Because there is no physical partition between zones, there could exist significant thermal impact among the adjacent zones than in other commercial buildings. It is critical to accurately analyze the energy interaction and reduce the modeling uncertainty for supporting advanced control. We propose a virtual partition surface temperature sensor for quantifying the variables and solve this challenge by using linear parametric models. The derived models based on physical law can be used as a guideline on selecting appropriate orders in mainly ARMAX (autoregressive moving average with exogenous variables) and/or ARX for improving the sensor’s performance. Validation of the virtual temperature models is conducted by three validation criteria: goodness of fit (G), mean squared error (MSE) and coefficient of determination (R2) under off-control conditions. The validation results show that the physical model based linear parametric model, ARX 211, performs well and similar to other system identification models, such as ARMAX 2111 and ARX 221, for estimating surface temperatures. The sensitivity analysis using three on-control conditions (under-sizing, proper-sizing and over-sizing condition) is conducted for analyzing and evaluating the performance and barriers of this virtual sensor. The proposed easy-to-implement model can be applied to support supervisory control of equipment in multi-zone buildings and other applications to supplement the measurements, like estimating the temperature of a structure integrated cooling or heating application in renewable energy areas.