基于负荷预测的冰蓄冷空调系统优化控制研究
|
摘要
为了更好地平衡电网负荷和缓解城市供电压力,冰蓄冷空调在我国得到了日益广泛的应用。如何最优化的利用冰蓄冷空调,充分发挥冰蓄冷空调移峰填谷的优势,使用户在经济上达到最大的利益,已经成为发展冰蓄冷空调的一个重要问题。近年来,优化控制策略已经逐渐取代了以往简单的单纯依靠主机或冰槽的控制策略,成为冰蓄冷空调系统的主要控制策略。
本文以广州大学城的外融冰盘管式冰蓄冷空调系统为研究对象,通过对如何实现冰蓄冷空调优化控制的研究,提出了冰蓄冷空调系统优化控制的一般过程,给出了进行负荷预测的方法以及负荷优化分配的数学模型。
本文首先介绍了广州大学城采用的冰蓄冷空调系统,包括大学城区域供冷系统、第二冷站冰蓄冷主要设备、运行工况和自控系统,并采集了此工程2007年5～7月份的室外温度和冷负荷数据,将这些数据作为建立负荷预测模型的依据。
其次,准确的逐时负荷预测是实施冰蓄冷系统优化控制的重要前提。本文根据人工神经网络原理,建立了室外温度预测的神经网络模型,使得24小时提前逐时温度预测平均相对误差从ASHRAE计算方法的5.21%降低到了0.71%;针对工作日和假日两种不同日期类型,分别建立了日总冷负荷的神经网络模型,对工作日日负荷预测的平均相对误差是4.71%,对假日日负荷预测为8.62%,这一结果基本满足实际应用的要求;在温度预测模型的基础上,进一步建立了空调提前24小时的逐时负荷ANN预测模型,计算该模型平均绝对误差为188Rth,平均相对误差为5.15%,期望相对误差为2.08%。
最后,建立了优化控制运行费用最小化的数学模型并进行了模型求解。求解结果表明,在日冷负荷小于蓄冰槽蓄冰总量且逐时冷负荷小于融冰供冷最大功率时,优化控制就是融冰单独供冷,蓄冰槽蓄冰量等于预测的日冷负荷再乘以一定的冗余系数。而在日冷负荷大于蓄冰总量,且有逐时负荷大于融冰供冷最大功率时,优化控制就是根据剩余冰量,在保持单台制冷机负荷系数0.7以上(一般为0.8)启动相应冷水机组台数,通过调节融冰供冷来满足时刻负荷需求,蓄冰槽冰量全部蓄满。
同时结合自控系统,在文章的最后一小节给出了冰蓄冷空调系统的一个优化控制方案设计,以此可以用来指导冰蓄冷空调控制的设计应用。
In order to better balance the electric supply and lighten the tension of urban power grid, ice storage air-conditioning is becoming more popular in our country. There is a key problem for developing ice storage system that how to take optimal advantage of the ice storage system, and make the ice storage system more effective. Throwing these, consumer can achieve the best benefit. During recent years, the optimal control has replaced the chiller priority and ice priority as the most important control method in the ice storage system gradually.
Base on the external-melt ice-on-coil thermal storage system in Guangzhou University City, we study of the optimal control of ice storage system and bring up the currently course of optimal control in ice storage system. In this course, we get the way how to predict the thermal load and the math model of load distribution.
Firstly, this paper introduce the ice storage air-condoning system in Guangzhou University City, including the district cooling system, main equipment, operating condition and BAS. Author also collected the outdoor temperature and load data in May to July 2007, which provided the sample data for load prediction model.
Accurate hourly load prediction is the important precondition for the ice storage system's optimal control. Based on BP theory, author establish temperature predicting ANN model for the next 24 hours. By this method, mean relative error (MRE) is reduced to 0.71% from 5.21%, compared to ASHRAE method. Then day cooling load predicting ANN model for workday and holiday is established respectively, and MRE reaches 4.71% for workday and 8.62% for holiday. According to the temperature predicting model has made, a next-24-hours hourly cooling load prediction ANN model is established, and mean absolute error (MAE) of hourly cooling load prediction is 188Rth. MRE is 5.15% and expected error percentage (EEP) is 2.08%.
Finally, a cost-minimum model for ice storage system is established and numerical calculation is carried out. When cooling load is less than ice-melting ability, optimal control strategy is just ice priority, capacity of ice is equivalent to the prediction cooling load multiplied by a certain redundancy coefficient. When cooling load is more than ice-melting ability, according to the remaining ice, optimal control is to keep single chiller load to half of chiller cooling ability (80% normally) and tune ice-melting to meet load, the ice-storage tank is full of ice.
Combining BAS, an optimal control blue print of ice storage air-conditioning is given in the final section, which can be used to guide the design of ice storage sysem.
本文以广州大学城的外融冰盘管式冰蓄冷空调系统为研究对象,通过对如何实现冰蓄冷空调优化控制的研究,提出了冰蓄冷空调系统优化控制的一般过程,给出了进行负荷预测的方法以及负荷优化分配的数学模型。
本文首先介绍了广州大学城采用的冰蓄冷空调系统,包括大学城区域供冷系统、第二冷站冰蓄冷主要设备、运行工况和自控系统,并采集了此工程2007年5～7月份的室外温度和冷负荷数据,将这些数据作为建立负荷预测模型的依据。
其次,准确的逐时负荷预测是实施冰蓄冷系统优化控制的重要前提。本文根据人工神经网络原理,建立了室外温度预测的神经网络模型,使得24小时提前逐时温度预测平均相对误差从ASHRAE计算方法的5.21%降低到了0.71%;针对工作日和假日两种不同日期类型,分别建立了日总冷负荷的神经网络模型,对工作日日负荷预测的平均相对误差是4.71%,对假日日负荷预测为8.62%,这一结果基本满足实际应用的要求;在温度预测模型的基础上,进一步建立了空调提前24小时的逐时负荷ANN预测模型,计算该模型平均绝对误差为188Rth,平均相对误差为5.15%,期望相对误差为2.08%。
最后,建立了优化控制运行费用最小化的数学模型并进行了模型求解。求解结果表明,在日冷负荷小于蓄冰槽蓄冰总量且逐时冷负荷小于融冰供冷最大功率时,优化控制就是融冰单独供冷,蓄冰槽蓄冰量等于预测的日冷负荷再乘以一定的冗余系数。而在日冷负荷大于蓄冰总量,且有逐时负荷大于融冰供冷最大功率时,优化控制就是根据剩余冰量,在保持单台制冷机负荷系数0.7以上(一般为0.8)启动相应冷水机组台数,通过调节融冰供冷来满足时刻负荷需求,蓄冰槽冰量全部蓄满。
同时结合自控系统,在文章的最后一小节给出了冰蓄冷空调系统的一个优化控制方案设计,以此可以用来指导冰蓄冷空调控制的设计应用。
In order to better balance the electric supply and lighten the tension of urban power grid, ice storage air-conditioning is becoming more popular in our country. There is a key problem for developing ice storage system that how to take optimal advantage of the ice storage system, and make the ice storage system more effective. Throwing these, consumer can achieve the best benefit. During recent years, the optimal control has replaced the chiller priority and ice priority as the most important control method in the ice storage system gradually.
Base on the external-melt ice-on-coil thermal storage system in Guangzhou University City, we study of the optimal control of ice storage system and bring up the currently course of optimal control in ice storage system. In this course, we get the way how to predict the thermal load and the math model of load distribution.
Firstly, this paper introduce the ice storage air-condoning system in Guangzhou University City, including the district cooling system, main equipment, operating condition and BAS. Author also collected the outdoor temperature and load data in May to July 2007, which provided the sample data for load prediction model.
Accurate hourly load prediction is the important precondition for the ice storage system's optimal control. Based on BP theory, author establish temperature predicting ANN model for the next 24 hours. By this method, mean relative error (MRE) is reduced to 0.71% from 5.21%, compared to ASHRAE method. Then day cooling load predicting ANN model for workday and holiday is established respectively, and MRE reaches 4.71% for workday and 8.62% for holiday. According to the temperature predicting model has made, a next-24-hours hourly cooling load prediction ANN model is established, and mean absolute error (MAE) of hourly cooling load prediction is 188Rth. MRE is 5.15% and expected error percentage (EEP) is 2.08%.
Finally, a cost-minimum model for ice storage system is established and numerical calculation is carried out. When cooling load is less than ice-melting ability, optimal control strategy is just ice priority, capacity of ice is equivalent to the prediction cooling load multiplied by a certain redundancy coefficient. When cooling load is more than ice-melting ability, according to the remaining ice, optimal control is to keep single chiller load to half of chiller cooling ability (80% normally) and tune ice-melting to meet load, the ice-storage tank is full of ice.
Combining BAS, an optimal control blue print of ice storage air-conditioning is given in the final section, which can be used to guide the design of ice storage sysem.
引文
[1]朱清宇.规范·引导·促进--《蓄冷空调工程技术规程》编制要点[J].上海建材,2006(1):24-25
[2]徐飞.浅谈冰蓄冷[J].智能建筑与城市信息,2007(3):35-37
[3]张永铨.我国蓄冷技术的现状及发展[C].中国制冷学会2007年学术年会论文集,2007
[4]邱东.广州大学城区域供冷系统[J].制冷空调与电力机械,2007,28(4):76-79
[5]周永山.冰蓄冷空调技术及其应用[J].资源节约与环保,2004(1):38-44
[6]龚少博.冰蓄冷系统的控制策略[J].智能建筑与城市信息,2007(3):32-34
[7]郭齐传,李育.冰蓄冷空调系统运行控制策略综述[J].建筑物理与设备,2003(1):41-52
[8]殷亮,刘道平,陈之航.蓄冷空调控制策略的选择及运行优化[J].能源研究与信息,1997,13(1):11-16
[9]佟威,阎秀英,介军.冰蓄冷系统优化控制策略及实时控制研究[J].暖通与设备,2007,35(10):38-44
[10]汪训昌.关于蓄冷空调工程设计思路的探讨[J].暖通空调 HV&AC,1998,28(3):20-23
[11]石磊,李兆东,刘伟.冰蓄冷空调系统模拟优化技术探综[J].建筑热能通风空调,2003(2):37-39
[12]D.H.Stethmann,Optimal control for cool storage.ASHRAE Transactions.1989.95(1):1189-1193
[13]刘业凤,史钟璋,张吉光,等.宾馆类建筑冰蓄冷空调系统优化控制的经济性分析[J].通风除尘,1998(3):21-24
[14]王勇,赵庆珠.冰蓄冷系统的优化控制分析[J].暖通空调,1996,26(3)
[15]广州大学城发展规划编制工作组.广州大学城规划咨询及发展规划综述[J].城市规划,2002,26(5)
[16]杨国良,林拥军.PLC在区域供冷自控系统中的应用及分析[J].电气传动自动 化,2006,28(3):49-52
[17]广东省建筑设计研究院.广州大学城区域供冷自控系统设计说明.2004
[18]杭州华电华源环境工程有限公司.广州大学城区域供冷冰蓄冷系统控制策略.2004
[19]Forrest,J.R.and W.J.Wepfer.Formulation of a load prediction algorithm for a large commercial building.ASHARE Transactions,90(2B):523-35,1984
[20]MacArthur J.W.,Mathur A.,Zhao J.On-line recursive estimation for load profile prediction.ASHRAE Transactions,1989,v95 n1:621:628
[21]Seem,J.E.and J.E.Braun.Aadaptive Methods for Real-time Forecasting of Building Electrical Demand.ASHRAE Transactions,97(1):710-21,1991
[22]Kreider,J.F.and X.A.Wang.Artificial neural networks demonstration for automated generation of energy use predictors for commercial buildings.ASHRAE Transactions,97(2):775-79,1991
[23]Gibson,G.L.and T.T.Kraft.Electric demand prediction using artificial neural network technology.ASHRAE Journal 35(3):60-68,1993
[24]Kawashima M.Artificial neural network back propagation model with three-phase annealing developed for the building energy predictor shoot-out.ASHRAE Transactions,1994,100(2):1096-1118
[25]Kawashima M.,Hourly thermal load prediction for the next 24 hours by ARIMA,EWMA,LR,and a neural network,ASHARE Transactions,1995,v101n1:186-200
[26]Kawashima M.,Dorgan C.E.,Mitchell J.W.Optimizing system control with load prediction by neural networks for an ice-storage system.ASHRAE Transactions,1996,v102 n1:1169-1178
[27]Sakawa M.,Ushiro S.,Kato K.,et al.Cooling load prediction in a district heating and cooling system through simplified robust filter and multi-layered neural network.Proceedings of the IEEE International Conference on Systems.Man and Cybenetics v3,1999,IEEE,USA.Ⅲ-995-Ⅲ-1000
[28]李元旦.建筑物空调负荷预测方法及冰蓄冷空调系统运行优化控制研究[D].上海:同济大学,2002.6
[29]何大四,张旭.改进的季节性指数平滑法预测空调负荷分析[J].同济大学学报,2005,33(12):1672-1676
[30]曹双华,曹家枞,李涛.基于小波变换的神经网络空调负荷预测研究[J].暖通空调HV,2005,35(4):13-17
[31]吴杰,陈光明.空调冷负荷的人工神经网络预测分析[C].中国工程热物理学会第十届年会,
[32]许冬,吴铮.基于MATLAB6.X的系统分析与设计-神经网络[M].陕西:西安电子科技大学出版社,2002.
[33]李国勇.智能控制及其MATLAB实现[M].北京:电子工业出版社,2006.
[34]郭军.BP神经网路算法研究[D].湖北:华中科技大学硕士学位论文,2005.
[35]韦钢,王飞,张永健,等.负荷预测中历史数据缺损处理[J].电力科学与工程,2004(1):16-19
[36]罗若谷,陈敏.BP神经网络泛化能力改进研究[J].福建电脑,2007(1):40-41
[37]ASHRAE.2001 ASHRAE Handbook-Fundamentals,Chapter 26,Table 2,Atlanta:American Society of Heating,Refrigerating and Air-conditioning Engineers,Inc.,2001
[38]常晓珂,夏清,晋欣桥.冰蓄冷空调系统控制中的负荷预测方法[J].建筑热能通风空调,2003(1):16-19
[39]石磊,赵蕾,王军,等.应用人工神经网络预测建筑物空调负荷[J].暖通空调,2003,33(1):103-104
[40]孙文瑜,徐成贤,朱德通.最优化方法[M].北京:高等教育出版社,2005.
[41]曹卫华,郭正.最优化技术方法及MATLAB的实现[M].北京:化学工业出版社,2005.
[2]徐飞.浅谈冰蓄冷[J].智能建筑与城市信息,2007(3):35-37
[3]张永铨.我国蓄冷技术的现状及发展[C].中国制冷学会2007年学术年会论文集,2007
[4]邱东.广州大学城区域供冷系统[J].制冷空调与电力机械,2007,28(4):76-79
[5]周永山.冰蓄冷空调技术及其应用[J].资源节约与环保,2004(1):38-44
[6]龚少博.冰蓄冷系统的控制策略[J].智能建筑与城市信息,2007(3):32-34
[7]郭齐传,李育.冰蓄冷空调系统运行控制策略综述[J].建筑物理与设备,2003(1):41-52
[8]殷亮,刘道平,陈之航.蓄冷空调控制策略的选择及运行优化[J].能源研究与信息,1997,13(1):11-16
[9]佟威,阎秀英,介军.冰蓄冷系统优化控制策略及实时控制研究[J].暖通与设备,2007,35(10):38-44
[10]汪训昌.关于蓄冷空调工程设计思路的探讨[J].暖通空调 HV&AC,1998,28(3):20-23
[11]石磊,李兆东,刘伟.冰蓄冷空调系统模拟优化技术探综[J].建筑热能通风空调,2003(2):37-39
[12]D.H.Stethmann,Optimal control for cool storage.ASHRAE Transactions.1989.95(1):1189-1193
[13]刘业凤,史钟璋,张吉光,等.宾馆类建筑冰蓄冷空调系统优化控制的经济性分析[J].通风除尘,1998(3):21-24
[14]王勇,赵庆珠.冰蓄冷系统的优化控制分析[J].暖通空调,1996,26(3)
[15]广州大学城发展规划编制工作组.广州大学城规划咨询及发展规划综述[J].城市规划,2002,26(5)
[16]杨国良,林拥军.PLC在区域供冷自控系统中的应用及分析[J].电气传动自动 化,2006,28(3):49-52
[17]广东省建筑设计研究院.广州大学城区域供冷自控系统设计说明.2004
[18]杭州华电华源环境工程有限公司.广州大学城区域供冷冰蓄冷系统控制策略.2004
[19]Forrest,J.R.and W.J.Wepfer.Formulation of a load prediction algorithm for a large commercial building.ASHARE Transactions,90(2B):523-35,1984
[20]MacArthur J.W.,Mathur A.,Zhao J.On-line recursive estimation for load profile prediction.ASHRAE Transactions,1989,v95 n1:621:628
[21]Seem,J.E.and J.E.Braun.Aadaptive Methods for Real-time Forecasting of Building Electrical Demand.ASHRAE Transactions,97(1):710-21,1991
[22]Kreider,J.F.and X.A.Wang.Artificial neural networks demonstration for automated generation of energy use predictors for commercial buildings.ASHRAE Transactions,97(2):775-79,1991
[23]Gibson,G.L.and T.T.Kraft.Electric demand prediction using artificial neural network technology.ASHRAE Journal 35(3):60-68,1993
[24]Kawashima M.Artificial neural network back propagation model with three-phase annealing developed for the building energy predictor shoot-out.ASHRAE Transactions,1994,100(2):1096-1118
[25]Kawashima M.,Hourly thermal load prediction for the next 24 hours by ARIMA,EWMA,LR,and a neural network,ASHARE Transactions,1995,v101n1:186-200
[26]Kawashima M.,Dorgan C.E.,Mitchell J.W.Optimizing system control with load prediction by neural networks for an ice-storage system.ASHRAE Transactions,1996,v102 n1:1169-1178
[27]Sakawa M.,Ushiro S.,Kato K.,et al.Cooling load prediction in a district heating and cooling system through simplified robust filter and multi-layered neural network.Proceedings of the IEEE International Conference on Systems.Man and Cybenetics v3,1999,IEEE,USA.Ⅲ-995-Ⅲ-1000
[28]李元旦.建筑物空调负荷预测方法及冰蓄冷空调系统运行优化控制研究[D].上海:同济大学,2002.6
[29]何大四,张旭.改进的季节性指数平滑法预测空调负荷分析[J].同济大学学报,2005,33(12):1672-1676
[30]曹双华,曹家枞,李涛.基于小波变换的神经网络空调负荷预测研究[J].暖通空调HV,2005,35(4):13-17
[31]吴杰,陈光明.空调冷负荷的人工神经网络预测分析[C].中国工程热物理学会第十届年会,
[32]许冬,吴铮.基于MATLAB6.X的系统分析与设计-神经网络[M].陕西:西安电子科技大学出版社,2002.
[33]李国勇.智能控制及其MATLAB实现[M].北京:电子工业出版社,2006.
[34]郭军.BP神经网路算法研究[D].湖北:华中科技大学硕士学位论文,2005.
[35]韦钢,王飞,张永健,等.负荷预测中历史数据缺损处理[J].电力科学与工程,2004(1):16-19
[36]罗若谷,陈敏.BP神经网络泛化能力改进研究[J].福建电脑,2007(1):40-41
[37]ASHRAE.2001 ASHRAE Handbook-Fundamentals,Chapter 26,Table 2,Atlanta:American Society of Heating,Refrigerating and Air-conditioning Engineers,Inc.,2001
[38]常晓珂,夏清,晋欣桥.冰蓄冷空调系统控制中的负荷预测方法[J].建筑热能通风空调,2003(1):16-19
[39]石磊,赵蕾,王军,等.应用人工神经网络预测建筑物空调负荷[J].暖通空调,2003,33(1):103-104
[40]孙文瑜,徐成贤,朱德通.最优化方法[M].北京:高等教育出版社,2005.
[41]曹卫华,郭正.最优化技术方法及MATLAB的实现[M].北京:化学工业出版社,2005.