基于GA-BP的NFT水培生菜根区温度预测
Root zone temperature prediction of NFT hydroponic lettuce based on GA-BP
刘艺梦 1潘守江 2丁小明 2王会强 1杜亚尊 3冯佐龙4
作者信息
- 1. 河北农业大学机电工程学院,河北保定,071001
- 2. 农业农村部规划设计研究院,北京市,100125
- 3. 河北省农业机械鉴定总站,石家庄市,050031
- 4. 河北省农机化技术推广总站,石家庄市,050000
- 折叠
摘要
营养液膜栽培技术(Nutrient Film Technique,NFT)模式下,作物对环境变化更加敏感.为保障作物根区环境条件合理,需要精准调控栽培管道内的温度,从而有效提高水培生菜品质,同时降低整体温室环境调控能耗.采用遗传算法(Genetic Algorithm,GA)优化BP神经网络模型的输入权重和阈值,以单个NFT栽培槽为研究对象,对槽内根区不同的监测区域分别构建温度预测模型,并与标准的BP神经网络和卷积神经网络(Convolutional Neural Network,CNN)模型进行对比.结果表明,GA-BP预测模型与标准BP和CNN神经网络模型相比,均方根误差分别降低0.82和0.42,平均绝对误差分别降低0.54和0.25,绝对系数分别提高0.08和0.03.该方法可提高基于BP神经网路算法对NFT根区温度预测模型精确度,为根区温度的精准控制提供可靠依据.
Abstract
Under the mode of Nutrient Film Technique(NFT),crops are more sensitive to environmental changes.In order to ensure reasonable environmental conditions in the root zone of crops,it is necessary to accurately regulate the temperature in the cultivation pipeline,so as to effectively improve the quality of hydroponic lettuce and reduce the overall energy consumption of greenhouse environmental regulation.In this study,Genetic Algorithm(GA)was used to optimize the input weights and thresholds of the BP neural network model.Taking a single NFT cultivation tank as the research object,temperature prediction models were constructed for different monitoring regions in the root zone of the tank.The Convolutional Neural Network(CNN)models are compared with the standard BP neural network and convolutional neural network.The results show that compared with the standard BP and CNN neural network models,the root mean square error of the GA-BP prediction model is reduced by 0.82 and 0.42,the average absolute error is reduced by 0.54 and 0.25,and the absolute coefficient is increased by 0.08 and 0.03,respectively.This method improves the accuracy of NFT root-zone temperature prediction model based on BP neural network algorithm and provides a reliable basis for the precise control of root-zone temperature.
关键词
水培生菜/营养液膜技术/根区温度预测/神经网络/遗传算法Key words
nutrient solution membrane technology/neural networks/root zone temperature prediction引用本文复制引用
基金项目
农业农村部规划设计研究院自主研发计划项目(QD202106)
设施园艺智能装备创新团队项目(CXTD-2021-05)
农规英才项目(QNYC-2021-02)
出版年
2024
国家科技期刊平台
NSTL
万方数据