摘要
通过分析影响甜菜产量的自然因素,选取6个主要影响因子应用于一种改进粒子群算法优化BP神经网络的预测模型.首先,在标准粒子群算法(Particle Swarm Optimization,PSO)中引入自适应惯性权重的方法增强搜索能力并且提高收敛速度,使用反向逃逸策略避免早熟现象的发生;将改进的粒子群算法引入到BP中形成NCPSO-BP的预测模型算法,既缩短了运算时间,又提高了预测精度;最后将NCPSO-BP与PSO-BP的预测效果进行对比,结果表明NCPSO-BP预测模型其最优预测结果的相对误差平均值3.59%,绝对误差平均值0.1969,比PSO-BP模型预测误差有所下降.通过这次智慧农业实验项目的应用,实现当年甜菜产量增产50%,为未来推广到面积更大、机械化程度更高的农田应用打下了基础,对现代化农业具有一定意义.
By analyzing the natural factors which affect beet yield,select six main influence factors and apply to an improved particle swarm algorithm to optimize the prediction model of BP neural network.First,an adaptive inertial weighting method is introduced in the standard particle swarm optimization(Particle Swarm Optimization,PSO)to enhance search capabilities and improve convergence speed;Use reverse escape strategy to avoid precocity.The improved particle swarm optimization algorithm is introduced into BP to form the prediction model algorithm of NCPSO-BP,the algorithm shortens the operation time and improves the prediction accuracy.Finally,the prediction effects of NCPSO-BP and PSO-BP are compared,the results show that relative error mean of the NCPSO-BP prediction model is 3.59%and mean absolute error is 0.1969,which is lower than that of the PSOBP model.Through this intelligent agricultural experiment project,a 50%increase in sugar beet production was achieved that year,laying a foundation for future extension to larger areas and deeper mechanized farmland,which has certain significance for modern agriculture.
作者
顾丽丽
刘勇
甄佳奇
GU Lili;LIU Yong;ZHEN Jiaqi(School of Electric Engineering,Heilongjiang University,Harbin Heilongjiang 150080,China)
出处
《新疆大学学报（自然科学版）（中英文）》
2021年第2期191-196,共6页
Journal of Xinjiang University(Natural Science Edition in Chinese and English)
基金
国家自然科学基金(61501176)
黑龙江省自然科学基金优秀青年项目(YQ2019F015)。
关键词
产量预测
粒子群算法
BP神经网络
动态权重
yield forecast
particle swarm optimization
BP neural network
dynamic weighting
