中文摘要：

传统GM（1，1）模型在参数α的绝对值较小的情况下近期刖电量预测精度较高，中期用电量预测往往误差较大，一定程度上是由于GM（1，1）模型的背景值x^（1）（k）取前后2个时刻的平均值造成的。引入向量θ得背景值序列的精确计算式，将GM（1，1）模型推广为GM（1，1，θ）模型。应用微粒群优化这一智能算法求解最优向鼠，从而构建GM（1，1，θ）模型，并将该模型应用于山东省中期用电量预测。实例分析表明，与传统GM（1，1）预测模型相比，智能优化模型较好地得到了预测点的预测结果，更适用于中期用电量预测。

英文摘要：

The conventional GM（ 1, 1 ） model is accurate in short-term electricity demand forecasting but has more errors in middle-term forecasting when the absolute value of the parameter a is relatively small. To some degree this is because that the background value x^（1） （k） of GM（ 1,1 ） is defined as the average of two sequent moments. The vector θ was introduced into the accurate calculation formula of background value array, and GM （ 1,1 ） was consequently generalized into GM （ 1,1,θ）. As particle swarm optimization has the virtue of optimum seeking, it was applied to solving the value of θ as well as the optimization model. The forecasting results demonstrate that the intelligent optimization model GM （1,1,θ） has higher forecast precision and adaptability for middle-term electricity demand forecasting.