针对电力负荷预测序列存在随机增长特性和非线性波动性,灰色神经网络模型能够有效地反映序列增长特性和对非线性关系进行拟合。引入灰色关联度作为确定组合预测权重的依据,以序列整体拟合优化为目标,从而找到能够代表序列内在变化规律性的组合权重,使总体预测精度得到提高。 In view of the stochastic growth and non-linear volatility of power load forecasting sequence, gray neural network model can effectively reflect the sequence growth characteristics and fit the nonlinear relationship. The gray relational degree is introduced as the basis to determine the combined forecasting weight, and the whole sequence fitting optimization is taken as the goal, so as to find the combination weight that can represent the inherent variation regularity of the sequence, and to improve the overall forecasting accuracy.