详细信息
基于GRNN的冰蓄冷空调逐时冷负荷预测 被引量:4
GRNN Based Hourly Cooling Load Prediction of Ice-storage Air-conditioning
文献类型:期刊文献
中文题名:基于GRNN的冰蓄冷空调逐时冷负荷预测
英文题名:GRNN Based Hourly Cooling Load Prediction of Ice-storage Air-conditioning
作者:徐今强[1,2,3];肖睿[1,2];董凯军[1];黄冲[1];何世辉[1];冯自平[1]
机构:[1]中国科学院广州能源研究所,广州510640;[2]中国科学院研究生院,北京100049;[3]湛江广东海洋大学信息学院,湛江524088
年份:2009
卷号:25
期号:1
起止页码:268
中文期刊名:微计算机信息
外文期刊名:Control & Automation
基金:国家自然科学基金委员会--广东省联合基金资助项目(U0634005)"高密度储能及潜热输送材料的基础物性及传输机理研究";国家863计划项目(2006AA05Z254)"高能效潜热输送关键技术研究及中试";广东省科技计划项目(2006A10705002)"动态冰蓄冷关键技术研究及示范"
语种:中文
中文关键词:冰蓄冷空调;负荷预测;义回归神经网络;传算法;滑因子
外文关键词:ice-storage air-conditioning; cooling load prediction; general regression neural network; genetic algorithm; smoothing factor
中文摘要:冷负荷动态预测对冰蓄冷空调最优化控制来说是不可或缺的。建立了基于广义回归神经网络(GRNN)和遗传算法(GA)的逐时冷负荷预测模型,建模时以前一日已知的24小时室外干球温度为输入,以次日逐时冷负荷为输出。为提高预测精度及改善鲁棒性,以均方差(MSE)最小构造适应度函数,应用遗传算法寻优广义回归神经网络的平滑因子。通过预测负荷与实际负荷的比较分析验证了模型的可靠性和鲁棒性。
外文摘要:Dynamic prediction of cooling load is indispensable to the optimal control of ice-storage air-conditioning. An hourly cooling load prediction model integrating general regression neural network (GRNN) and genetic algorithm (GA) was presented. The external hourly dry bulb temperature recorded every hour between 1:00 and 24:00 of the previous day were used as network input, the hourly cooling load for the next day as output in the model. In order to improve the accuracy and robustness of the model, a fitness function aiming at the mean square error (MSE) was constructed and the GA was employed to optimize smoothing factors of the GRNN. The reliability and robustness of this model were verified by comparing the predicted load with the actual load.
参考文献:
正在载入数据...