[1]杜佳敏,李舒宏,李新美.基于ResNet-GWO的冷源系统节能优化[J].东南大学学报(自然科学版),2020,50(5):866-874.[doi:10.3969/j.issn.1001-0505.2020.05.011]
 Du Jiamin,Li Shuhong,Li Xinmei.Energy-saving optimization of cooling system based on ResNet-GWO[J].Journal of Southeast University (Natural Science Edition),2020,50(5):866-874.[doi:10.3969/j.issn.1001-0505.2020.05.011]
点击复制

基于ResNet-GWO的冷源系统节能优化()
分享到:

《东南大学学报(自然科学版)》[ISSN:1001-0505/CN:32-1178/N]

卷:
50
期数:
2020年第5期
页码:
866-874
栏目:
能源与动力工程
出版日期:
2020-09-20

文章信息/Info

Title:
Energy-saving optimization of cooling system based on ResNet-GWO
作者:
杜佳敏1李舒宏1李新美2
1东南大学能源与环境学院, 南京 210096; 2南京福加自动化科技有限公司, 南京 210046
Author(s):
Du Jiamin1 Li Shuhong1 Li Xinmei2
1School of Energy and Environment, Southeast University, Nanjing 210096, China
2Nanjing Fuca Automation Technology Co., Ltd., Nanjing 210046, China
关键词:
中央空调冷源系统 节能运行 残差神经网络 灰狼优化
Keywords:
central air-conditioning system energy-saving operation residual neural network grey wolf optimizer(GWO)
分类号:
TU831.3
DOI:
10.3969/j.issn.1001-0505.2020.05.011
摘要:
针对中央空调冷源系统运行能耗高、设备之间高度耦合机理建模困难、参数众多难以随环境变化动态调节的问题,引入智能化算法进行建模优化.以某经过初步节能改造且冷水主机并联运行的地铁站空调冷源系统实测运行数据为基础,结合冷源系统运行原理建立能耗预测残差神经网络(ResNet)模型. 采用灰狼优化算法(GWO)对某典型夏季制冷日运行工况进行寻优计算.仿真结果表明,冷源系统ResNet模型在测试集上的平均相对误差、平均绝对误差值分别为1.548 5%、2.239 4,预测精度高于BP神经网络模型和支持向量回归机模型;优化结果显示,GWO优化后的能耗值相比遗传算法(GA)、粒子群优化(PSO)更低,较实际运行平均节能10.45%,其中冷水主机能耗降低8.14%,而各主机冷冻水供水温度相等时冷机节能率仅为5.37%.因此,基于ResNet-GWO的仿真优化策略可用于实现中央空调冷源系统的高能效运行.
Abstract:
To solve the problem that the central air-conditioning system parameters are difficult to dynamically adjust with the environment and cause high energy consumption, an intelligent algorithm for modeling and optimization was introduced. Based on the measured operating data of an air-conditioning cooling system with chillers operating in parallel in a subway station by preliminary energy-saving transformation, a ResNet(residual neural network)energy consumption model was established according to the operating characteristics. Combined with the ResNet model, a grey wolf optimizer(GWO)was used to optimize the operating parameters of operational conditions in central air-conditioning cooling system at one summer day. The verification results show that the ResNet model has an average relative prediction error of 1.548 5% and the mean absolute error of 2.239 4. The prediction accuracies of ResNet model are higher than that of BP neural network model and support vector regression(SVR)model. The optimization results show that the optimized energy consumption of GWO is lower than that of the genetic algorithm(GA)and the particle swarm optimization(PSO), and the simulation energy consumption after optimization decreases 10.45% on average, which the energy consumption of the chillers is reduced by 8.14%, while the energy saving rate of the chillers is only 5.37% when the chilled water supply temperature of each chiller is equal. Thus, therefore, the proposed simulation and optimization strategy based on ResNet-GWO can be used to realize high energy efficiency operation of cooling systems.

参考文献/References:

[1] Pérez-Lombard L, Ortiz J, Pout C. A review onbuildings energy consumption information[J]. Energy and Buildings, 2008, 40(3): 394-398. DOI:10.1016/j.enbuild.2007.03.007.
[2] Li F F, Zhang Z P, Yu J. Study of public building energy consumption status and energy saving potential[J]. Applied Mechanics and Materials, 2014, 641/642: 982-986. DOI:10.4028/www.scientific.net/amm.641-642.982.
[3] Tashtoush B, Molhim M, Al-Rousan M. Dynamic model of an HVAC system for control analysis[J]. Energy, 2005, 30(10): 1729-1745. DOI:10.1016/j.energy.2004.10.004.
[4] Lu L, Cai W J, Chai Y S, et al. Global optimization for overall HVAC systems: Part Ⅰ problem formulation and analysis[J]. Energy Conversion and Management, 2005, 46(7/8): 999-1014. DOI:10.1016/j.enconman.2004.06.012.
[5] Chen C L, Chang Y C, Chan T S. Applying smart models for energy saving in optimal chiller loading[J]. Energy and Buildings, 2014, 68: 364-371. DOI:10.1016/j.enbuild.2013.04.030.
[6] Nabil N. Modeling and optimization of HVAC systems using artificial neural network and genetic algorithm[J]. Building Simulation, 2014, 7(3): 237-245. DOI:10.1007/s12273-013-0138-3.
[7] He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Las Vegas, NV, USA:IEEE, 2016: 770-778. DOI:10.1109/CVPR.2016.90.
[8] Yu J Q, Liu Q T, Zhao A J, et al. Optimal chiller loading in HVAC system using a novel algorithm based on the distributed framework[J]. Journal of Building Engineering, 2020, 28: 101044. DOI:10.1016/j.jobe.2019.101044.
[9] Lu L, Cai W J, Soh Y C, et al. Global optimization for overall HVAC systems: Part Ⅱ problem solution and simulations[J]. Energy Conversion and Management, 2005, 46(7/8): 1015-1028. DOI:10.1016/j.enconman.2004.06.011.
[10] Kusiak A, Li M Y, Tang F. Modeling and optimization of HVAC energy consumption[J]. Applied Energy, 2010, 87(10): 3092-3102. DOI:10.1016/j.apenergy.2010.04.008.
[11] Lee W S, Chen Y T, Kao Y C. Optimal chiller loading by differential evolution algorithm for reducing energy consumption[J]. Energy and Buildings, 2011, 43(2/3): 599-604. DOI:10.1016/j.enbuild.2010.10.028.
[12] Mirjalili S, Mirjalili S M, Lewis A. Grey wolf optimizer[J]. Advances in Engineering Software, 2014, 69: 46-61. DOI:10.1016/j.advengsoft.2013.12.007.
[13] Friedman J H. Greedy function approximation: A gradient boosting machine[J]. Annals of Statistics, 2001, 29(5): 1189-1232. DOI:10.1214/aos/1013203451.
[14] Benesty J, Chen J D, Huang Y T. On the importance of the Pearson correlation coefficient in noise reduction[J]. IEEE Transactions on Audio, Speech, and Language Processing, 2008, 16(4): 757-765. DOI:10.1109/TASL.2008.919072.
[15] Ioffe S, Szegedy C. Batch normalization: Accelerating deep network training by reducing internal covariate shift[EB/OL].(2015)[2020-04]. https://arxiv.org/abs/1502.03167.
[16] Xu B, Wang N Y, Chen T Q, et al. Empirical evaluation of rectified activations in convolutional network[EB/OL].(2015)[2020-04].https://arxiv.org/abs/1505.00853.
[17] Wang H D. Water flow rate models based on the pipe resistance and pressure difference in multiple parallel chiller systems[J]. Energy and Buildings, 2014, 75: 181-188. DOI:10.1016/j.enbuild.2014.02.017.
[18] 刘雪峰,郑宇蓝,王家绪,等.多台冷水机组非均匀负荷匹配运行控制策略[J].华南理工大学学报(自然科学版),2019,47(9):24-32,39. DOI:10.12141/j.issn.1000-565X.190031.
Liu X F, Zheng Y L, Wang J X, et al. Operational control strategy of non-uniform load matching for multiple chillers[J]. Journal of South China University of Technology(Natural Science Edition), 2019, 47(9): 24-32, 39. DOI:10.12141/j.issn.1000-565X.190031. (in Chinese)
[19] 曾振威,杨伟才,李聪.部分负荷工况下多台冷水机组的冷量分配[J].暖通空调,2008,38(3):107-110. DOI:10.3969/j.issn.1002-8501.2008.03.026.
Zeng Z W, Yang W C, Li C. Output distribution of a multiple chiller system under part load conditions[J]. Heating Ventilating & Air Conditioning, 2008, 38(3): 107-110. DOI:10.3969/j.issn.1002-8501.2008.03.026. (in Chinese)
[20] 刘金平,周登锦.空调系统变冷水温度调节的节能分析[J].暖通空调,2004,34(5):90-91,96. DOI:10.3969/j.issn.1002-8501.2004.05.021.
Liu J P, Zhou D J. Energy efficiency analysis of variable chilled water temperature adjustment in air conditioning systems[J]. Heating Ventilating & Air Conditioning, 2004, 34(5): 90-91, 96. DOI:10.3969/j.issn.1002-8501.2004.05.021. (in Chinese)

备注/Memo

备注/Memo:
收稿日期: 2020-04-22.
作者简介: 杜佳敏(1996—),女,硕士生;李舒宏(联系人),男,博士,教授,equart@seu.edu.cn.
基金项目: 国家重点研发计划资助项目(2017YFC0702501).
引用本文: 杜佳敏,李舒宏,李新美.基于ResNet-GWO的冷源系统节能优化[J].东南大学学报(自然科学版),2020,50(5):866-874. DOI:10.3969/j.issn.1001-0505.2020.05.011.
更新日期/Last Update: 2020-09-20