基于奇异谱分析和极限学习机的风速多步预测

风速预测对风力发电系统具有重要的影响，为获得更高精度的风速预测结果，针对多步风速预测，成功开发了一种基于奇异谱分析和优化极限学习机的新型预测模型。首先，采用奇异谱分析将风速时间序列分解为一组相对平稳的分量，以降低风速序列的随机性对预测结果的影响；然后，对分解得到的分量分别建立极限学习机预测模型，为进一步提高预测性能，将1种新颖的活性竞争萤火虫算法用于优化极限学习机的输入权值和隐含层偏置；最后，叠加全部分量的预测值得到实际预测结果。仿真结果表明，基于奇异谱分析和活性竞争萤火虫算法优化极限学习机的模型在1步到3步风速预测中实现了较高精度的预测结果。

Multi-step wind speed forecasting based on singular spectrum analysis and extreme learning machine

Wind speed forecasting is an essential influence on wind power systems. In order to obtain more accurate wind speed prediction results,a new forecasting model based on singular spectrum analysis (SSA) and optimized extreme learning machine (ELM) is successfully developed for multi-step wind speed forecasting. Firstly, SSA is used to decompose the wind speed sequence into a group of relatively smooth components to reduce the impact of its randomness on the predicted results.Secondly forecasting models based on the ELMs were established respectively for each component, to further improve the forecasting performance, a novel active competitive firefly algorithm (ACFA) is used to optimize the input weights and bias of the hidden nodes of ELM. Finally, the prediction results of each component were superimposed to obtain the final wind speed forecasting results. The simulation results showed that the proposed model based on the singular spectrum analysis and extreme learning machine optimized by active competitive firefly algorithm (SSA-ACFA-ELM) achieved higher accurate forecasting results in one-step to three-step wind speed prediction.