基于CEEMD与VHBFO_SVM的微网短期负荷预测模型

Microgrid Short-Term Load Forecasting Model Based on CEEMD and VHBFO_SVM

为适应微网的建设和发展对其负荷预测效率及精度的要求,提出一种基于变概率混合细菌觅食优化算法(VHBFO)优化支持向量机(SVM)的微网短期负荷预测模型。首先利用CEEMD将非平稳的负荷序列按照不同波动尺度逐级进行分解,从而得到多组固有模态函数分量均值,并建立VHBFO_SVM模型对各组分量分别进行预测,最后通过叠加各组分量的预测结果得到预测值。以国内某微网示范工程项目为例,将VHBFO_SVM用于微网短期负荷预测。实例仿真结果表明,所提出的VHBFO_SVM预测模型优于SVM预测模型,更适用于当前微网短期负荷预测需要。

In order to meet the requirement of the efficiency and accuracy of load forecasting for the construction and development of microgrid, a short-term load forecasting model based on variable probability hybrid bacterial foraging optimization algorithm (VHBFO) and optimized support vector machine (SVM) for microgrid has been proposed. Firstly, the non-stationary load series have been decomposed step by step according to different fluctuation scales by using CEEMD, and the mean values of components of multiple groups of intrinsic modal functions have been obtained. The VHBFO_SVM model has been established to predict each component separately. Finally, the prediction value has been obtained by superimposing the prediction results of each component. Taking a microgrid demonstration project in China as an example, VHBFO_SVM has been applied to the short-term load forecasting of microgrid. The simulation results show that the proposed VHBFO_SVM forecasting model has better performance than the SVM forecasting model, and is more suitable for short-term load forecasting of microgrid.