基于神经网络的温室光环境调控模型改进研究

Research on improved regulatory model of greenhouse light environment based on neural network

温室种植在我国应用很普遍,主要原因是温室的保温性能得到了广大种植户的青睐。然而温室在不同的光照条件下室内外温差较大,由于光照引起的温室种植损害事件屡有发生,因此根据光照对温室进行科学控制与搭建有着重要的意义。文章首先根据实验基地的大棚结构,设计4种模型,根据温室自身结构特征分析不同模型下温室地温和气温的变化特征,基于不同距离光照强度对比与不同温室内连续三天气温日变化比较,引入激励函数的离散值,通过误差纠正学习使得输出值趋于最优化,根据仿真曲线,可以准确判断温室光环境的强度及影响,输出时结果稳定性接近于正常状态,具有很强的实用性。

Greenhouse cultivation is widely used in China mainly for the greenhouse is insulated by the majority of growers favor.However,the temperature is large different indoor and outdoor in different illumination conditions.Crop damage incidents have occurred for the light inducing in greenhouse.It has an important significance to control and set up the greenhouse according to the illumination.Firstly,according to the structure of the greenhouse in experiment base,it designed 4models,and constructed the control model of greenhouse light environment according to the structure characteristics of greenhouse itself combined with neural network weight value of neural PID controller.different distances and different light intensity contrast in the greenhouse for three consecutive days and the temperature daily variation based on discrete into comparison,incentive function,through error correction learning makes the output value to optimization,based on the simulation curve,which can accurately determine the strength and influence of greenhouse light environment,and output stability results is close to the normal state,which has a strong practical.