摘要
利用标准BP神经网络建立短期电力负荷预测模型,其算法存在最终解过于依赖初值和过学习现象,并且训练过程中存在局部极小问题且预测精度低等缺点。为了提高电力负荷模型的预测精度,通过阅读相关文献,构建了基于改进BP神经网络的短期电力负荷预测模型,该模型采用遗传算法对权值和阈值进行初始化,以相对误差和附加动量法相结合的方式去计算权值修正量。比较改进后的BP算法和标准BP算法在短期电力负荷预测的效果,从实验仿真的效果表明改进后的模型提高了预测精度。
The short-term power load forecasting is established based on standard BP neural network model. There are phenomenons such as the final solution relying too much on initial value and the overfitting, and the algorithm has disadvantages such as local minimum problems and slow convergence speed and so on in the training process. In order to improve the prediction accuracy of power load model, through reading literature,the model constructs the short-term power load forecasting based on improved BP neural network model, uses genetic algorithm to initialize weights and threshold values, and uses relative error and the way of additional momentum method to calculate weight correction. Comparing the effect of the improved BP algorithm with the standard BP algorithm in the short-term power load forecasting, the simulation experiment results show that the improved model can improve the prediction precision.
出处
《微型机与应用》
2017年第14期61-63,67,共4页
Microcomputer & Its Applications
基金
中央高校基本科研业务费专项资金项目(DL11AB01)
关键词
短期负荷预测
BP神经网络
遗传算法
相对误差
附加动量
short-term load forecasting
BP neural network
genetic algorithm
relative error
additional momentum