针对影响风电中长期预测的气象、地理等因素众多且复杂,及无法解决长期依赖时间序列的问题,提出一种基于多维特征融合网络(multi-dimensional feature fusion network,MFFN)和长短期记忆(long and short term memory,LSTM)的预测方法—...针对影响风电中长期预测的气象、地理等因素众多且复杂,及无法解决长期依赖时间序列的问题,提出一种基于多维特征融合网络(multi-dimensional feature fusion network,MFFN)和长短期记忆(long and short term memory,LSTM)的预测方法—多维特征提取(feature extraction,FE)-关联函数(copula,CO)-LSTM融合模型(FE-CO-LSTM)。收集来自不同地区4个风电场的特征数据,在研究云贵高原地区风电场的背景下,最大限度扩充模型数据集;使用关联结构函数构造一种提取气象特征的方法,使模型可以在一定程度上量化气象因素和风力发电之间的相关性;基于神经网络模型提出一种特征表示与融合方法,以有效表达风电场气象因素、地理位置等特征;最后提出一种基于LSTM网络的中长期发电量预测模型,以有效解决模型对风电场时间序列数据反向传播时早期月度数据信息缺失的问题。实验结果证明,FE-CO-LSTM表现出最佳的预测性能。展开更多
China wind atlas was made by numerical simulation and the wind energy potential in China was calculated. The model system for wind energy resource assessment was set up based on Canadian Wind Energy Simulating Toolkit...China wind atlas was made by numerical simulation and the wind energy potential in China was calculated. The model system for wind energy resource assessment was set up based on Canadian Wind Energy Simulating Toolkit (WEST) and the simulating method was as follows. First, the weather classes were obtained depend on meteorological data of 30 years. Then, driven by the initial meteorological field produced by each weather class, the meso-scale model ran for the distribution of wind energy resources according each weather class condition one by one. Finally, averaging all the modeling output weighted by the occurrence frequency of each weather class, the annual mean distribution of wind energy resources was worked out. Compared the simulated wind energy potential with other results from several activities and studies for wind energy resource assessment, it is found that the simulated wind energy potential in mainland of China is 3 times that from the second and the third investigations for wind energy resources by CMA, and is similar to the wind energy potential obtained by NREL in Solar and Wind Energy Resource Assessment(SWERA) project. The simulated offshore wind energy potential of China seems smaller than the true value. According to the simulated results of CMA and considering lots of limited factors to wind energy development, the final conclusion can be obtained that the wind energy availability in China is 700～1 200 GW, in which 600～1 000 GW is in mainland and 100～200 GW is on offshore, and wind power will become the important part of energy composition in future.展开更多
Leveraging the commercial CFD software FLUENT,the fine-scale three-dimensional wind structure over the Paiya Mountains on the Dapeng Peninsula near Shenzhen,a city on the seashore of South China Sea,during the landfal...Leveraging the commercial CFD software FLUENT,the fine-scale three-dimensional wind structure over the Paiya Mountains on the Dapeng Peninsula near Shenzhen,a city on the seashore of South China Sea,during the landfall of Typhoon Molave has been simulated and analyzed.Through the study,a conceptual wind structure model for mountainous areas under strong wind condition is established and the following conclusions are obtained as follows:(1)FLUENT can reasonably simulate a three-dimensional wind structure over mountainous areas under strong wind conditions;(2)the kinetic effect of a mountain can intensify wind speed in the windward side of the mountain and the area over the mountain peak;and(3)in the leeward side of the mountain,wind speed is relatively lower with relatively stronger wind shear and turbulence.展开更多
文摘针对影响风电中长期预测的气象、地理等因素众多且复杂,及无法解决长期依赖时间序列的问题,提出一种基于多维特征融合网络(multi-dimensional feature fusion network,MFFN)和长短期记忆(long and short term memory,LSTM)的预测方法—多维特征提取(feature extraction,FE)-关联函数(copula,CO)-LSTM融合模型(FE-CO-LSTM)。收集来自不同地区4个风电场的特征数据,在研究云贵高原地区风电场的背景下,最大限度扩充模型数据集;使用关联结构函数构造一种提取气象特征的方法,使模型可以在一定程度上量化气象因素和风力发电之间的相关性;基于神经网络模型提出一种特征表示与融合方法,以有效表达风电场气象因素、地理位置等特征;最后提出一种基于LSTM网络的中长期发电量预测模型,以有效解决模型对风电场时间序列数据反向传播时早期月度数据信息缺失的问题。实验结果证明,FE-CO-LSTM表现出最佳的预测性能。
文摘China wind atlas was made by numerical simulation and the wind energy potential in China was calculated. The model system for wind energy resource assessment was set up based on Canadian Wind Energy Simulating Toolkit (WEST) and the simulating method was as follows. First, the weather classes were obtained depend on meteorological data of 30 years. Then, driven by the initial meteorological field produced by each weather class, the meso-scale model ran for the distribution of wind energy resources according each weather class condition one by one. Finally, averaging all the modeling output weighted by the occurrence frequency of each weather class, the annual mean distribution of wind energy resources was worked out. Compared the simulated wind energy potential with other results from several activities and studies for wind energy resource assessment, it is found that the simulated wind energy potential in mainland of China is 3 times that from the second and the third investigations for wind energy resources by CMA, and is similar to the wind energy potential obtained by NREL in Solar and Wind Energy Resource Assessment(SWERA) project. The simulated offshore wind energy potential of China seems smaller than the true value. According to the simulated results of CMA and considering lots of limited factors to wind energy development, the final conclusion can be obtained that the wind energy availability in China is 700～1 200 GW, in which 600～1 000 GW is in mainland and 100～200 GW is on offshore, and wind power will become the important part of energy composition in future.
基金National Natural Science Foundation of China(91215302,51278308)Open Project for State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry,Institute of Atmospheric Physics(LAPC)
文摘Leveraging the commercial CFD software FLUENT,the fine-scale three-dimensional wind structure over the Paiya Mountains on the Dapeng Peninsula near Shenzhen,a city on the seashore of South China Sea,during the landfall of Typhoon Molave has been simulated and analyzed.Through the study,a conceptual wind structure model for mountainous areas under strong wind condition is established and the following conclusions are obtained as follows:(1)FLUENT can reasonably simulate a three-dimensional wind structure over mountainous areas under strong wind conditions;(2)the kinetic effect of a mountain can intensify wind speed in the windward side of the mountain and the area over the mountain peak;and(3)in the leeward side of the mountain,wind speed is relatively lower with relatively stronger wind shear and turbulence.
