科技通报
科技通報
과기통보
BULLETIN OF SCIENCE AND TECHNOLOGY
2013年
8期
10-12
,共3页
预测%最小二乘法%广义加权组合法%非线性规划
預測%最小二乘法%廣義加權組閤法%非線性規劃
예측%최소이승법%엄의가권조합법%비선성규화
prediction methods%least-squares method%the combined forecasting model based on generalized weighted proportional means%nonlinear programming
研究数据预测问题，传统最小二乘法所确定的单项预测模型，一方面不能很好地利用不同模型的信息，另一方面当拟合效果不佳时预测误差较大。针对最小二乘法的缺点，构建了一种新的预测模型---广义加权最小二乘组合预测模型。通过最小二乘法确立多个单项预测模型，经过广义加权组合法综合不同模型的信息，利用非线性规划法求解最优权系数，最终得到组合预测模型。经仿真计算，证明了该方法的可行性及优越性。
研究數據預測問題，傳統最小二乘法所確定的單項預測模型，一方麵不能很好地利用不同模型的信息，另一方麵噹擬閤效果不佳時預測誤差較大。針對最小二乘法的缺點，構建瞭一種新的預測模型---廣義加權最小二乘組閤預測模型。通過最小二乘法確立多箇單項預測模型，經過廣義加權組閤法綜閤不同模型的信息，利用非線性規劃法求解最優權繫數，最終得到組閤預測模型。經倣真計算，證明瞭該方法的可行性及優越性。
연구수거예측문제，전통최소이승법소학정적단항예측모형，일방면불능흔호지이용불동모형적신식，령일방면당의합효과불가시예측오차교대。침대최소이승법적결점，구건료일충신적예측모형---엄의가권최소이승조합예측모형。통과최소이승법학립다개단항예측모형，경과엄의가권조합법종합불동모형적신식，이용비선성규화법구해최우권계수，최종득도조합예측모형。경방진계산，증명료해방법적가행성급우월성。
Studying about the data forecasting problem. The least-squares method can’t make a full use of the informa-tion of different models, and will reduce the forecast precision since the fitting results is not ideal. According to the weakness, a new forecasting model based on generalized weighted proportional means and least-squares method is estab-lished. Firstly, the model used the least-squares method to determine the single forecast models. Secondly, the model adopted the generalized weighted proportional means to syncretize different informations from the single. Thirdly, the model estimate optional weight coefficients through nonlinear programming and the combined forecasting model can be get.The results of simulation demonstrated that the combined forecasting model was better than a single method.
