基于自回归和神经网络算法加权组合的负荷预测

Load Forecasting Based on Weighted Combination between Autoregression and Neural Network Algorithm

针对时间序列法的自回归动平均模型和神经网络算法在负荷预测中的不足,提出对这两种预测结果采用加权组合方法,在不同时期的负荷预测采用不同的加权值来提高预测结果的精确度。通过算例分析短期负荷预测和长期负荷预测,证明采用加权组合方法的预测结果比自回归动平均模型和神经网络算法分别预测要准确。

In view of disadvantages of autoregressive integrated moving average model and neural network algorithm of time series method in load forecasting,the paper proposes that the two forecasting results should adopt weighted combination method and load forecasting in different stages should adopt different weights to promote accuracy of forecasting results.Short-term load forecasting and long-term forecasting are analyzed via examples,which proves that forecasting result from weighted combination method is more accurate than that from either autoregressive integrated moving average model or neural network algorithm.