Artificial neural network analysis based on genetic algorithm to predict the performance characteristics of a cross flow cooling tower

International Conference on Energy Engineering and Environmental Protection 2017

能源学;生态环境科学

Wu, Jiasheng^1,2 ; Cao, Lin^1,3 ; Zhang, Guoqiang^4

School of Energy and Power Engineering, Nanjing University of Science and Technology, Nanjing

210094, China^1

Hunan Yuanheng Technology Co., Ltd, Changsha

410116, China^2

Postdoctoral Work Station, Guangdong Jirong Air Conditioning Equipment Co., Ltd., Jieyang

522000, China^3

College of Civil Engineering, Hunan University, Changsha

410082, China^4

关键词: Absorbing efficiency;    BP neural network model;    Ga-bp neural networks;    Genetic-algorithm optimizations;    Heat and mass transfer;    Non-linear relationships;    Performance characteristics;    Performance parameters;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/121/5/052001/pdf
DOI  :  10.1088/1755-1315/121/5/052001

学科分类：环境科学（综合）

来源: IOP

【 摘 要 】

Cooling tower of air conditioning has been widely used as cooling equipment, and there will be broad application prospect if it can be reversibly used as heat source under heat pump heating operation condition. In view of the complex non-linear relationship of each parameter in the process of heat and mass transfer inside tower, In this paper, the BP neural network model based on genetic algorithm optimization (GABP neural network model) is established for the reverse use of cross flow cooling tower. The model adopts the structure of 6 inputs, 13 hidden nodes and 8 outputs. With this model, the outlet air dry bulb temperature, wet bulb temperature, water temperature, heat, sensible heat ratio and heat absorbing efficiency, Lewis number, a total of 8 the proportion of main performance parameters were predicted. Furthermore, the established network model is used to predict the water temperature and heat absorption of the tower at different inlet temperatures. The mean relative error MRE between BP predicted value and experimental value are 4.47%, 3.63%, 2.38%, 3.71%, 6.35%,3.14%, 13.95% and 6.80% respectively; the mean relative error MRE between GABP predicted value and experimental value are 2.66%, 3.04%, 2.27%, 3.02%, 6.89%, 3.17%, 11.50% and 6.57% respectively. The results show that the prediction results of GABP network model are better than that of BP network model; the simulation results are basically consistent with the actual situation. The GABP network model can well predict the heat and mass transfer performance of the cross flow cooling tower.