Research on the Performance of Feed Forward Neural Network Based Temperature Field Identification Model in Intelligent Building
The identification of temperature filed of monitored region is one of the key steps for the energy efficiency management in intelligent building. In this paper, the identification of temperature field in monitored region is formalized as one optimization problem. With the formalization, a feed forward neural network is used to identify the temperature field of monitored region in an intelligent building. To improve the performance of the identification model, input data of the desired neural network is normalized with minimum-maximum method as middle result and the normalization image of the input data is the stereographic projection of the middle result. To test the performance of our proposed identification model, temperature matrix for the infrared photograph is used in our experiment. BP and RBF neural network is used as the desired neural network. Experiment results show that the performance of BP based temperature field identification model running with data preprocessed by stereographic projection and minimum-maximum method for the identification of temperature field in monitored region is better.
temperature field identification neural network stereographic projection
Zhenya Zhang Hongmei Cheng Shuguang Zhang
Key Laboratory of Intelligent Building of Anhui Province,Anhui University of Architecture Hefei,Chin School of Management,Anhui University of Architecture Hefei,China School of Management,University of Science and Technology of China Hefei,China
国际会议
上海
英文
219-223
2011-03-11（万方平台首次上网日期，不代表论文的发表时间）