农机化研究
農機化研究
농궤화연구
JOURNAL OF AGRICULTURAL MECHANIZATION RESEARCH
2015年
4期
230-236
,共7页
农机总动力%灰色模型%多元线性回归模型%Shapley值%组合预测
農機總動力%灰色模型%多元線性迴歸模型%Shapley值%組閤預測
농궤총동력%회색모형%다원선성회귀모형%Shapley치%조합예측
agricultural machinery total power%gray model%multiple linear regression model%Shapley value%combined prediction
为能获得精确预测农机总动力的方法,以灰色模型和多元线性回归模型为子模型,应用 Shapley 值法计算子模型权重系数,构建农机总动力组合预测模型。应用我国2000-2010年农机总动力数据,分别标定上述模型相关参数,并计算各模型年度相对误差和平均相对误差。其中, GM模型和多元线性回归模型的平均相对误差分别为0.68%和0.91%,组合预测模型的平均相对误差为0.59%,精度较高。同时,组合模型既能够反映数据自身变化规律的特征,又能定量反映农机总动力与其相关影响因数间的数理关系,具有较强的适用性。
為能穫得精確預測農機總動力的方法,以灰色模型和多元線性迴歸模型為子模型,應用 Shapley 值法計算子模型權重繫數,構建農機總動力組閤預測模型。應用我國2000-2010年農機總動力數據,分彆標定上述模型相關參數,併計算各模型年度相對誤差和平均相對誤差。其中, GM模型和多元線性迴歸模型的平均相對誤差分彆為0.68%和0.91%,組閤預測模型的平均相對誤差為0.59%,精度較高。同時,組閤模型既能夠反映數據自身變化規律的特徵,又能定量反映農機總動力與其相關影響因數間的數理關繫,具有較彊的適用性。
위능획득정학예측농궤총동력적방법,이회색모형화다원선성회귀모형위자모형,응용 Shapley 치법계산자모형권중계수,구건농궤총동력조합예측모형。응용아국2000-2010년농궤총동력수거,분별표정상술모형상관삼수,병계산각모형년도상대오차화평균상대오차。기중, GM모형화다원선성회귀모형적평균상대오차분별위0.68%화0.91%,조합예측모형적평균상대오차위0.59%,정도교고。동시,조합모형기능구반영수거자신변화규률적특정,우능정량반영농궤총동력여기상관영향인수간적수리관계,구유교강적괄용성。
In order to obtain accurate method of predicting Agricultural Machinery Total Power , Gray Model and Multiple Linear Regression Model were set to the sub-model , and Shapley Value was applied to calculate the weighting factors of sub-model , then the Combined Prediction Model of Agricultural Machinery Total Power was built .Application of Agri-cultural Machinery Total Power data in China from 2000 to 2010 , these models were calibrated parameters , then the rela-tive errors for each year and the average relative errors were calculated , where the average relative errors of gray model and Multiple Linear Regression Model are 0 .68% and 0 .91%, while the average relative error of Combined Prediction Model is 0 .59%, has high precision .And the Combined Prediction Model not only reflects the variation characteristics of the data itself , but also quantitatively reflects the mathematical relationship of Agricultural Machinery Total Power and its factors, has strong applicability .