基于实数遗传算法的农机总动力变权组合预测  

Variable Weight Combination Prediction of Total Power of Agricultural Machinery Based on Real-coded Genetic Algorithm

作　　者：陈信新 王福林[1] 宋莹莹 Chen Xinxin;Wang Fulin;Song Yingying(College of Engineering,Northeast Agricultural of University,Harbin 150030,China)

机构地区：[1]东北农业大学工程学院

出　　处：《农机化研究》2020年第1期1-7,42,共8页Journal of Agricultural Mechanization Research

基　　金：公益性行业(农业)科研专项(201503116-04)

摘　　要：黑龙江省农机总动力进行预测,选用指数函数模型、多项式拟合模型、三次指数平滑模型、龚帕兹曲线模型为单一预测模型并以1980-2013年农机总动力数据为样本点进行拟合,其平均绝对百分比误差分别为12.5%、3.22%、3.43%、6.09%;然后,建立以误差平方和最小为目标函数的变权组合预测模型,并利用改进的实数遗传算法对变权组合预测模型进行优化,所得到的变权重组合预测模型的平均绝对百分比误差为1.98%,拟合效果较好。以2014-2016年农机总动力数据为预测时点对该方法进行验证,结果表明:不论是拟合精度还是预测误差都具有较好的效果。最后,对黑龙江省未来5年的农机总动力进行了预测,为相关部门对农业机械化的发展规划提供参考。In order to predict the total power of agricultural machinery in Heilongjiang Province.First,the paper used exponential model,polynomial fitting model,three exponential smoothing model and gompertz curve model as single prediction model and used the data from 1980 to 2013 as sample points for fitting.The average absolute percentage error were 12.5%,3.22%,3.43%、6.09%.Then built a model of variable weight combination prediction with the objective function is minimum error sum of squares and used improved real-coded genetic algorithm to optimize the combination prediction model.The average absolute percentage error of variable weight combination prediction model is 1.98%,the fitting effect is good.This method was verified with the data from 2014 to 2016 as the prediction time points,the results show that both the fitting accuracy and the prediction error had good effects.Finally,it forecasted the total power of agricultural machinery for the next five years in Heilongjiang province,and it’s a reference to the development of agricultural mechanization of the relevant departments.

关 键 词：实数遗传算法 变异算子 组合预测 农机总动力 

分 类 号：S220.2[农业科学—农业机械化工程]