论文部分内容阅读
为克服利用传统最小二乘法估计GM(1,1)模型参数的缺陷,改善该模型在中长期负荷预测中的精度,提出了基于最小一乘法的GM(1,1)模型,对离散后的GM(1,1)模型以误差的绝对值之和达到最小确定参数。由于该GM(1,1)模型中的目标函数非连续,不可导,用传统的优化无法求解,本文针对模型的特性设计了求解该优化模型的遗传算法,对模型的参数进行估计。将所提出的模型应用于负荷预测,通过与传统预测效果的对比分析,验证了本文方法的有效性和优越性。
In order to overcome the shortcomings of estimating GM (1,1) model parameters by traditional least square method and improve the accuracy of the model in medium and long term load forecasting, a GM (1,1) model based on least square method is proposed. The GM (1,1) model reaches the minimum deterministic parameter with the sum of absolute errors. Because the objective function in the GM (1,1) model is not continuous and can not be induced, it can not be solved by traditional optimization. In this paper, a genetic algorithm to solve the optimization model is designed and the parameters of the model are estimated. The proposed model is applied to load forecasting, and compared with the traditional forecasting results to verify the effectiveness and superiority of the proposed method.