An Optimization Model for the Identification of Temperature in Intelligent Building

Introduction

With the rapid development of modern information technology on computer, communication and automation (Braun, 2007; Luo, Lin, Chen, & Su, 2006; Yang & Peng, 2001; Deo, 2006), intelligent building (Flax, 1991; Ralegaonkar & Gupta, 2010; Chen, Clements-Croome, Hong, Li, & Xu, 2006) is presented to describe the phenomena that more and more new building, especially new large public building are equipped with more and more smart equipments (Krukowski & Arsenijevic, 2010; Seo, Oh, Suh, & Park, 2007; Sazonov, Janoyan, & Jha, 2004; Hagras, 2008; Osterlind, Pramsten, Roberthson, Eriksson, Finne, & Voigt, 2007). Because there are many automated equipment in an intelligent building, the energy consumed in an intelligent building is cared by user and researcher. Because the distributing of temperature and it’s variety in intelligent building are related to the energy efficiency and comfort grade of the building closely, many researchers put their attention on temperature filed in intelligent building. If the temperature field in an intelligent building can be forecasted rightly, the process of energy consuming of the building can be interfered for higher energy efficiency while the comfort grade is kept better level on. To reconstruct temperature field with observation data of discrete observation position in intelligent space, more and more researchers are focused on the identification of temperature in intelligent building.

Although the identification of temperature field in a building is researched widely (Carmody & O'Mahony, 2009; Li, Qin, & Yue, 2008; Jiménez & Madsen, 2008; Zhang, 2009; Jiang, Mahadevan, & Adeli, 2007; Malti, Victor, & Oustaloup, 2008) and some valuable models are gained, most of those models are depended strictly on some constrains such as specific architecture structure, specific heat source as incentive environment. It is difficult and costly to gather those parameters related to temperature field in a building because the architecture of the building may be variety and there are many kinds of building equipment as heat source in intelligent building. For most building equipments can be treated as heat source and the architecture of that building may be variety, it is difficult to collect values of those parameters in real time. To value those parameters with time requirement satisfied, temperature of some observation positions in an intelligent building can be sampled in long term time and the temperature field in the building can be constructed according to the analysis result of those observation data (Lu, Miao, & He, 2009; Yu, 2006).

Identification structure and parameter estimating methods are two key factors of temperature identification (Xiao, Bai, & Yu, 2006; Zhong & Song, 2008). Because temperature in an intelligent building is influenced by many complicated parameters (Wang & Xu, 2005; Hassan, Guirguis, Shaalan, & El-Shazly, 2007), it is practical to assume that all of those parameters are stable in recent time. With this assumption, it is right to construct the identification model of temperature field in intelligent building with high frequency observation data of temperature in that building (Yu, Yi, & Zhao, 2008). To forecast the temperature near an observation position in a building, an optimization model for the reconstruction of temperature in an intelligent building and a method for the temperature identification in that building with feed forward neural network as identification structure and genetic algorithm for the parameter optimization of the method are presented in this paper. The rest of this paper is organized as following. The identification structure based on feed forward neural network and the optimization of identification parameters based on genetic algorithm are presented. Experimental results are shown with observation data of temperature in the electronic reading room of south laboratory in Anhui Architecture University.