基于CEEMD和机器学习算法的短期风速组合预测

Short-term Wind Speed Combination Prediction Based on CEEMD and Machine Learning Algorithm

风速的准确预测对电网的稳定和电力系统的安全运行非常重要。针对风速的非线性和非平稳性特征,提出基于完备总体经验模态分解(CEEMD)结合神经网络和最小二乘支持向量机(LSSVM)的短期风速组合预测模型。利用CEEMD将原始风速时间序列分解成有限个特征互异的模态分量;高频分量利用组合神经网络进行模拟预测,中低频分量使用LSSVM构建预测模型;对每一子序列预测结果进行重构,使用灰狼优化算法对权重矩阵实时调整,确定组合模型的最优权系数,得到最终预测值。经对国内某风电场进行实验,结果表明:组合风速预测模型具有较好的预测能力,在短期风速预测方面的可行性与有效性。

The accurate prediction of wind speed is very important to the stability of the power grid and the safe operation of the power system. Aiming at the nonlinear and non-stationary characteristics of wind speed, a short-term wind speed combined forecasting model based on complete total empirical mode decomposition(CEEMD) combined with neural network and least square support vector machine(LSSVM) is proposed. Firstly, use CEEMD to decompose the original wind speed time series into a limited number of modal components with different characteristics;Secondly, the high-frequency components are simulated and predicted by a combined neural network, and the mid-and low-frequency components are constructed using LSSVM to establish a prediction model;Finally, for each sub-component the sequence prediction results are reconstructed, the gray wolf optimization algorithm is used to adjust the weight matrix in real time, the optimal weight coefficient of the combined model is determined, and the final prediction value is obtained. Experiments on a domestic wind farm show that the proposed combined wind speed prediction model has good prediction capabilities, which verifies the feasibility and effectiveness of the combined model in short-term wind speed prediction.