改进的遗传灰色RBF模型的短期电力负荷预测  

A Gray RBF Model Improved by Genetic Algorithm for Short-Term Load Forecasting

作　　者：张毅[1] 姜思博[1] 李铮[1] 

机构地区：[1]东北电力大学经济管理学院,吉林吉林132012

出　　处：《电测与仪表》2014年第5期1-4,10,共5页Electrical Measurement & Instrumentation

摘　　要：为了在智能电网环境下提高短期电力负荷预测的精度,提出了一种考虑实时电价影响的遗传算法改进的灰色RBF模型。该方法利用灰色模型可以减弱数据随机性以及RBF神经网络的高度非线性的优点,弱化实时电价对短期电力负荷预测的影响,针对两种方法结合容易陷入局部最优和收敛性问题,采用遗传算法对网络进行了优化,得到最终预测结果。实例验证表明,与灰色RBF预测方法相比,该方法具有更高的负荷预测精度和较强的适应能力。In order to improve short-term load forecasting accuracy of the smart grid, this paper proposes a gray RBF model improved by genetic algorithm, which considers the influence of real-time electricity pricing. This method de-clines the influence of real-time pricing on short-term load forecasting using the characteristics that gray model can weaken data randomness and RBF neural network is highly nonlinear. Considering the problem of local optimum and convergence, the paper uses genetic algorithm for net optimization to obtain the final prediction result. Experiment shows that this method has higher precision and stronger adaptation ability than gray RBF model prediction.

关 键 词：智能电网 短期负荷预测 实时电价 遗传算法 神经网络 

分 类 号：TM715[电气工程—电力系统及自动化]