基于VMD-SE-CEEMDAN-PSO-SVR短期风速预测

Short-Term Wind Speed Forecasting Based on VMD-SE-CEEMDAN-PSO-SVR

摘要: 风能具有随机性和间歇性，准确可靠的风速预测对于风电场规划和电网运营规划至关重要｡本文提出VMD (variational mode decomposition, VMD)-SE (sample entropy, SE)-CEEMDAN (complete ensemble empirical mode decomposition with adaptive noise, CEEMDAN)-PSO (particle swarm optimization, PSO)-SVR (support vector regression, SVR)的组合分解短期风速预测｡首先是对VMD分解的子序列(intrinsic mode function, IMF)用样本熵(SE)判别方法对复杂度较高的子序列用自适应噪声的完备经验模态分解(CEEMDAN)进行二次分解，然后将两者分解后得到的序列用支持向量机(SVR)进行预测｡此外，为了找到更优的SVR参数，文中引入粒子群优化算法进行改进，并基于粒子群算法对SVR参数寻优，进而对某风电场进行短期风速预测，实验结果证明，VMD-SE-CEEMDAN-PSO-SVR模型的预测精度相对于其他模型更高｡

Abstract: Wind energy is random and intermittent, and accurate and reliable wind speed predictions are essential for wind farm planning and grid operation planning. This article presents: VMD (variational mode decomposition, VMD)-SE (sample entropy, SE)-CEEMDAN (complete ensemble empirical mode decomposition with adaptive noise, CEEMDAN)-PSO (particle swarm optimization, PSO)-SVR (support vector regression, SVR) is a combination of decomposition short-term wind speed prediction. Firstly, the intrinsic mode function (IMF) of VMD decomposition is decomposed by sample entropy (SE) discrimination method, and the subsequence with high complexity is decomposed by CEEMDAN, and then the sequence obtained by decomposition is predicted by support vector machine (SVR).In addition, in order to find better SVR parameters, the particle swarm optimization algorithm is introduced to improve it, and the SVR parameters are optimized based on the particle swarm optimization algorithm, and then the short-term wind speed prediction of a wind farm is carried out, and the experimental results show that the prediction accuracy of the VMD-SE-CEEMDAN-PSO-SVR model is higher than that other models.