【摘要】 针对新建风场面临缺乏历史数据且数据不完整的情况，文中考虑风场的数值天气预报（numerical weather forecast,NWP）中风速与风向、风场地形、风机尾流、风机湍流，对风场输出功率展开物理预测。鉴于物理预测精度存在一定局限性，故采用基于变分模态分解（variational modal decomposition,VMD）-蜣螂优化（dung beetle optimization,DBO）算法-深度极限学习机（deep extreme learning machine,DELM）优化组合方法 VMD-DBO-DELM进行残差修正。通过VMD方法将原始信号分解为多个模态函数，然后使用DBO算法优化DELM的参数，最后将优化后的DELM用于残差修正，有效提高了预测精度。通过与其他传统残差修正方法对比，VMD-DBO-DELM组合方法进一步提高了预测精度，为新建风场的功率预测提供了更为优越的解决方案和思路，在理论与实际应用方面都具有重要价值，为解决新建风场面临的数据问题及提高预测精度提供了可行途径。更多还原

【Abstract】 In view of the lack of historical data and incomplete data of the new wind farm,this paper considers the numerical weather forecast(wind speed,wind direction),wind field terrain,wind turbine wake and wind turbine turbulence to carry out physical prediction for the output power of the wind farm.However,in view of the limited accuracy of physical prediction,the optimization combination method(VMD-DBO-DELM),which is based on variable modal decomposition(VMD),dung beetle optimization(DBO),and deep extreme learning machine(DELM),is used to correct the residual error.The original signal is decomposed into multiple modal functions by the VMD method,and then the parameters of DELM are optimized by the DBO algorithm.Finally,the optimized DELM is used for residual correction,so as to improve the prediction accuracy.Compared with other traditional residual correction methods,the VMD-DBO-DELM combination method further improves the prediction accuracy.It provides a superior solution and ideas for the power prediction of new wind farms,which has important value in theory and practical application,and provides a feasible way to solve the data problems faced by new wind farms and improve the prediction accuracy.更多还原