改进两阶段分解的熵变混合短期风速预测研究

Prediction Study on Entropy Change Short-Term Wind Speed Based on Improved Two-Stage Decomposition

针对风速序列不平稳难以预测的问题,提出了一种混沌麻雀搜索算法(CSSA)优化最小二乘支持向量机(LSSVM)参数的短期风速预测混合模型。模型结合样本熵(SE)和具有自适应噪声改进的互补集成经验模态分解(ICEEMDAN)、变分模态分解(VMD)两阶段分解的数据预处理方法。首先,利用ICEEMDAN分解原始风速序列,且依据SE评估子序列的复杂程度,重构熵值近似的序列,VMD二次分解熵值最大的序列。然后对所有子序列分别建立LSSVM预测模型,同时CSSA对该模型参数优化以提高预测效率。最后将预测的各子序列叠加得到最终风速预测值。通过与经典模态分解等混合模型比较表明,所提基于优化算法的模型预测精度和收敛速度有明显提高。

Aiming at the problem that the wind speed series is unstable and difficult to predict, a hybrid model for short-term wind speed prediction based on chaotic sparrow search algorithm(CSSA) and optimizing the parameters of least squares support vector machine(LSSVM) is proposed. The model combines sample entropy(SE) and two-stage decomposition of complementary integrated empirical mode decomposition(iceemdan) and variational mode decomposition(VMD) with adaptive noise improvement.. First, the historical wind speed was decomposed with ICEEMDAN,then the approximate sample entrapy of sequence was reconstructed, and the decomposed maximum entrapy of sequence was combined with VMD according to SE quantitative analysis of subsequences complexity. Next, the least squares support vector machine forecasting model was established for all subsequences by decomposition and the optimization characteristics of CSSA algorithm were used to optimize the parameters of LSSVM to build a short-term wind power forecasting model. Finally, each of the forecasted subsequences was superposed to get the final wind speed forecast value. Compared with the classical model, the decomposition hybrid model shows that the forecasting accuracy and convergence speed of the proposed model based on the optimization algorithm are obviously improved.