| 利用相空间重构短时间间隔的吸收太阳能预测方法 |
| |
| 引用本文： | 汪心怡,姚彦鑫,董未名.利用相空间重构短时间间隔的吸收太阳能预测方法[J].实验室研究与探索,2020(1):32-36,47. |
| |
| 作者姓名： | 汪心怡 姚彦鑫 董未名 |
| |
| 作者单位： | ;1.北京信息科技大学光电测试技术及仪器教育部重点实验室;2.高端装备智能感知与控制北京市国际科技合作基地;3.中国科学院自动化研究所 |
| |
| 基金项目： | 北京市自然科学基金项目(4172021);北京市属高等学校高层次人才引进与培养计划项目(CIT&TCD201704064);国家重点研究培育项目(5221824108,5211910957);北京信息科技大学2019年度‘实培计划’项目。 |
| |
| 摘 要： | 考虑太阳能吸收能量的混沌特性和递归小波网络学习能力强、动态适应性强的优点,提出采用相空间重构与递归小波网络结合的预测模型。分别用该模型与未进行相空间重构、普通递归网络等模型分析实测数据比较,发现采用相空间重构后数据进行模型训练与预测对各种天气、环境的适应性强;而且预测间隔越短,计算量降低的效果越明显;当预测间隔为1.25 min时,计算量降低25%以上。此外,该模型具有相对简单、需要历史数据少、存储空间少的优点,而且预测精度高。
|
| 关 键 词： | 递归小波网络 相空间重构 能量吸收 |
Harvested Solar Energy Short Term Prediction Method Utilizing Phase Space Reconstruction |
| |
| WANG Xinyi,YAO Yanxin,DONG Weiming.Harvested Solar Energy Short Term Prediction Method Utilizing Phase Space Reconstruction[J].Laboratory Research and Exploration,2020(1):32-36,47. |
| |
| Authors: | WANG Xinyi YAO Yanxin DONG Weiming |
| |
| Institution: | (Key Laboratory of the Ministry of Education for Optoelectronic Measurement Technology and Instrument,Advanced Equipment Intelligent Perception and Control,Beijing International Cooperation Base for Science and Technology,Beijing Information Science&Technology University,Beijing 100010,China;Institute of Automation Chinese Academy of Sciences,Beijing 100080,China) |
| |
| Abstract: | In this paper,considering the chaotic characteristics of solar energy harvesting,and the advantages of strong learning ability and strong dynamic adaptability,a prediction model combining phase space reconstruction and recurrent wavelet network is proposed.The measured data are analyzed by the proposed method and the models of the non-phase space reconstruction and the ordinary recursive network.It is found that because the proposed model is trained and predicted by the phase space reconstruction data,the adaptability to various weather and environment is strong.Moreover,the shorter is the prediction interval,the more obvious is the effect of reducing calculation cost.When the prediction interval is 1.25 min,the computation amount is reduced by more than 25%.In addition,the model is relatively simple with less historical data and less storage space requirement.However,the prediction accuracy is high. |
| |
| Keywords: | recurrent wavelet network phase space reconstruction energy harvesting |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
