熵判别人工蜂群算法优化的风电功率组合预测模型

Combined Wind Power Prediction Model Optimized by Entropy Criterion Artificial Bee Colony Algorithm

为了提高风电功率预测精度,文中提出一种熵判别人工蜂群算法优化的组合模型。该模型是以最小相对误差作为目标函数,通过熵判别人工蜂群算法选取最优折现因子,确定模型权重系数,进而提高模型性能;熵判别人工蜂群算法通过计算蜜蜂的熵值,调节种群的多样性,对蜂群的搜索进行动态权重调整,同时对部分适应度值较差的蜜蜂进行迁移,增强蜜蜂的动态搜索能力。实验表明:熵判别人工蜂群算法优化的组合模型,能够智能地确定权重系数,较其他常规组合模型其预测精度明显提高。

In order to improve the prediction precision of wind power, a combined prediction model optimized by entro- py criterion artificial bee colony （ECABC） algorithm is proposed. With the minimum relative error as an objective func- tion, the optimal discount factor can be selected by using ECABC algorithm, and weight coefficients can be determined to further improve the model performance. The proposed algorithm can adjust the diversity of the population and dynami- cally adjust the weights of bees＇ searching by computing the entropy of bees. In the meantime, the bees that have inferi- or fitness values are moved to improve the searching capability dynamically. Experiments show that the proposed com- bined model can determine the weight coefficients smartly; and compared with other traditional combined models, the prediction precision is improved obviously.