风能因其安全、可再生、环保等显著优势而受到世界各国的重视,为了准确预测风速时间序列,本文使用宁夏回族自治区麻黄山共17,376条风速数据,提出了一种基于集合经验模态分解(EEMD)、长短期记忆网络(LSTM)、鲸鱼优化算法(WOA)组成的混合...风能因其安全、可再生、环保等显著优势而受到世界各国的重视,为了准确预测风速时间序列,本文使用宁夏回族自治区麻黄山共17,376条风速数据,提出了一种基于集合经验模态分解(EEMD)、长短期记忆网络(LSTM)、鲸鱼优化算法(WOA)组成的混合风速预测模型,并且与BP神经网络、CEEMDAN-LSTM-PSO (完全集合经验模态分解–长短期记忆网络–粒子群优化算法)、EMD-LSTM-RIME (经验模态分解–长短期记忆网络–霜冰优化算法)等模型进行对比实验,结果表明本文提出的EEMD-LSTM-WOA模型有着更稳定、更准确的预测性能。之后对EEMD-LSTM-WOA模型进行消融试验,结果显示去掉EEMD分解后,RMSE和MAPE分别增加了203.97%和187.47%,表明EEMD极大提升了整个模型的准确性和稳定性;去掉鲸鱼优化算法后,模型的RMSE和MAPE分别增加了78.34%和74.93%,说明最优化方法对整个模型的准确性和稳定性也有较大的促进作用。Wind energy has garnered global attention due to its notable advantages, including safety, renewability, and environmental friendliness. To accurately predict wind speed time series, this paper utilizes 17,376 wind speed data points from Ma Huang Mountain in the Ningxia Hui Autonomous Region. We propose a hybrid wind speed prediction model that combines ensemble empirical mode decomposition (EEMD), long short-term memory network (LSTM), and whale optimization algorithm (WOA). Comparative experiments were conducted with models such as BP neural network, CEEMDAN-LSTM-PSO (complete ensemble empirical mode decomposition-long short-term memory network-particle swarm optimization), and EMD-LSTM-RIME (empirical mode decomposition-long short-term memory network-Rimoglio optimization algorithm). The results indicate that our proposed EEMD-LSTM-WOA model exhibits more stable and accurate prediction performance. Subsequently, ablation experiments were performed on the EEMD-LSTM-WOA model. The findings revealed that upon removing EEMD decomposition, RMSE and MAPE increased by 203.97% and 187.47%, respectively, highlighting the significant enhancement of EEMD in boosting the model’s accuracy and stability. Similarly, after eliminating the whale optimization algorithm, the RMSE and MAPE of the model rose by 78.34% and 74.93%, respectively, indicating that this optimization method significantly contributes to the model's accuracy and stability.展开更多
针对传统电压源逆变器无模型预测电流控制(model-free predictive current control,MFPCC)方法存在电流纹波大、电流梯度更新停滞以及预测性能易受采样扰动影响的问题。该文提出一种计及采样扰动的三矢量MFPCC方法。在一个控制周期应用...针对传统电压源逆变器无模型预测电流控制(model-free predictive current control,MFPCC)方法存在电流纹波大、电流梯度更新停滞以及预测性能易受采样扰动影响的问题。该文提出一种计及采样扰动的三矢量MFPCC方法。在一个控制周期应用3个基本矢量,并根据价值函数计算矢量作用时间,降低了输出电流纹波;其次,通过建立不同矢量作用下的电流梯度方程组,实现电流梯度数据的实时更新,消除了停滞现象;再次,分析采样扰动对MFPCC的影响,采用扩张状态观测器估计采样扰动以补偿预测电流控制,抑制其对输出电流的影响。最后,通过仿真和实验,对所提方法的有效性进行了验证。展开更多
文摘风能因其安全、可再生、环保等显著优势而受到世界各国的重视,为了准确预测风速时间序列,本文使用宁夏回族自治区麻黄山共17,376条风速数据,提出了一种基于集合经验模态分解(EEMD)、长短期记忆网络(LSTM)、鲸鱼优化算法(WOA)组成的混合风速预测模型,并且与BP神经网络、CEEMDAN-LSTM-PSO (完全集合经验模态分解–长短期记忆网络–粒子群优化算法)、EMD-LSTM-RIME (经验模态分解–长短期记忆网络–霜冰优化算法)等模型进行对比实验,结果表明本文提出的EEMD-LSTM-WOA模型有着更稳定、更准确的预测性能。之后对EEMD-LSTM-WOA模型进行消融试验,结果显示去掉EEMD分解后,RMSE和MAPE分别增加了203.97%和187.47%,表明EEMD极大提升了整个模型的准确性和稳定性;去掉鲸鱼优化算法后,模型的RMSE和MAPE分别增加了78.34%和74.93%,说明最优化方法对整个模型的准确性和稳定性也有较大的促进作用。Wind energy has garnered global attention due to its notable advantages, including safety, renewability, and environmental friendliness. To accurately predict wind speed time series, this paper utilizes 17,376 wind speed data points from Ma Huang Mountain in the Ningxia Hui Autonomous Region. We propose a hybrid wind speed prediction model that combines ensemble empirical mode decomposition (EEMD), long short-term memory network (LSTM), and whale optimization algorithm (WOA). Comparative experiments were conducted with models such as BP neural network, CEEMDAN-LSTM-PSO (complete ensemble empirical mode decomposition-long short-term memory network-particle swarm optimization), and EMD-LSTM-RIME (empirical mode decomposition-long short-term memory network-Rimoglio optimization algorithm). The results indicate that our proposed EEMD-LSTM-WOA model exhibits more stable and accurate prediction performance. Subsequently, ablation experiments were performed on the EEMD-LSTM-WOA model. The findings revealed that upon removing EEMD decomposition, RMSE and MAPE increased by 203.97% and 187.47%, respectively, highlighting the significant enhancement of EEMD in boosting the model’s accuracy and stability. Similarly, after eliminating the whale optimization algorithm, the RMSE and MAPE of the model rose by 78.34% and 74.93%, respectively, indicating that this optimization method significantly contributes to the model's accuracy and stability.
文摘针对传统电压源逆变器无模型预测电流控制(model-free predictive current control,MFPCC)方法存在电流纹波大、电流梯度更新停滞以及预测性能易受采样扰动影响的问题。该文提出一种计及采样扰动的三矢量MFPCC方法。在一个控制周期应用3个基本矢量,并根据价值函数计算矢量作用时间,降低了输出电流纹波;其次,通过建立不同矢量作用下的电流梯度方程组,实现电流梯度数据的实时更新,消除了停滞现象;再次,分析采样扰动对MFPCC的影响,采用扩张状态观测器估计采样扰动以补偿预测电流控制,抑制其对输出电流的影响。最后,通过仿真和实验,对所提方法的有效性进行了验证。
