Research On The Short-term Forecast Model Of The Temperature Of The Northern Sunshine Panel Greenhouse Based On Meteorological Data

In order to realize the planting and production of crops throughout the year,China has vigorously developed advanced equipment and technologies applied to facility agriculture.Among them,the modern greenhouse is an important part of facility agriculture,which can create a good environment suitable for the growth of various crops.Greenhouse temperature is one of the most important factors affecting crop growth.How to accurately predict greenhouse temperature is one of the hot spots in facility agriculture research at this stage.In the greenhouse temperature prediction model currently constructed,on the one hand,the temperature change is considered from the perspective of mechanism,the model construction is complicated,a large number of parameters are difficult to determine,and the accuracy is relatively low.On the other hand,the method of system identification is selected,and most of them use neural network algorithm to establish the input and output model of greenhouse temperature changes.The input of the model is mostly the environmental variables inside and outside the greenhouse and the state variables of the equipment,which is rarely combined with the local weather forecast,and there are few short-term prediction models to study the temperature of the northern solar panel greenhouse.Therefore,in order to reasonably introduce meteorological data into the greenhouse temperature short-term prediction model,this paper first constructs an outdoor temperature prediction model based on meteorological data and other parameters,and uses its output as one of the input of the prediction model,and then builds an LM-BP neural network greenhouse Temperature short-term model.Finally,the model is optimized by mind evolution algorithm to improve the prediction effect of greenhouse temperature.The main research content includes the following aspects:(1)Based on the energy conservation theory,construct the temperature mechanism model of the solar panel test greenhouse and determine the environmental variables to be collected in the test.And the proposed BP neural network algorithm is studied from the aspects of network structure and principle,which provides a certain theoretical basis for the subsequent greenhouse temperature modeling.(2)The Delphi software is used to compile the greenhouse environment collection hostcomputer to realize real-time collection of related environmental variables inside and outside the greenhouse,and the cluster analysis method using the correlation analysis of matter-element analysis is used to scientifically and reasonably control the temperature sensor.At the same time,after filling in the missing relevant data inside and outside the greenhouse,Kalman filter processing,normalization processing and correlation analysis are performed in sequence.(3)Construct a short-term prediction model of greenhouse temperature based on meteorological data.In order to ensure that a small number of effective input variables are selected when constructing the model,combined with the results of correlation analysis,new ideas are used to more reasonably introduce meteorological forecast data to build a short-term model of greenhouse temperature,that is,first use meteorological data and outdoor environment related variables at time t Temperature prediction model,and then its output and other related variables as input,greenhouse temperature as output,to build a short-term temperature model of the greenhouse.Build short-term LM-BP temperature prediction models that predict the next moment and the next 12 hours.At the same time,construct a short-term LM-BP temperature prediction model that directly takes meteorological data and other relevant variables as input.Through the analysis of the test results,the use of new ideas to introduce meteorological forecast data to construct the experimental greenhouse short-term temperature model has better prediction effect,and can better predict the temperature change trend in the next moment and short-term12-hour greenhouse.(4)The mind evolution algorithm(MEA)was used to optimize the temperature prediction model LM-BP and the short-term 12-hour temperature prediction model LM-BP12 at the next moment.At the same time,a genetic algorithm(GA)was used to establish a short-term greenhouse temperature prediction model of GA-LM-BP neural network.By comparing the evaluation indexes between the various models,the root mean square errors of the MEA-LM-BP model and the MEA-LM-BP12 model are 1.194? and 2.583?,respectively,and the model validity is 99.01% and 96.14%,respectively.The BP model is superior to other models in terms of prediction effect and model validity.This study combines meteorological data with facility agriculture,and realizes the prediction of the greenhouse temperature at the next moment and the next 12 hours through theneural network algorithm,and uses the thinking evolution algorithm to optimize the model,providing a theoretical basis for the reasonable regulation of greenhouse temperature.