Study On Temperature And Humidity Prediction And Early Warning System In Greenhouse Based On SVM

Intelligent agriculture has always occupied a huge proportion in the intelligent economy,whether developed or developing countries,intelligent agriculture is its main way to consolidate the economic base,highlight the advantages of latecomer,and achieve catch-up strategy.Therefore,it is imperative to constantly explore more advanced and practical methods of intelligent agriculture.As one of the application fields of intelligent agriculture,the prediction and early warning of temperature and humidity in agricultural greenhouse is to realize the collection of agricultural field environmental data by using Internet of things equipment such as sensors,and to use mathematics and related statistical learning methods.This paper aims at the problems existing in the prediction and early warning of temperature and humidity in agricultural greenhouse.,the first is to realize serial port communication through nodemcu and CC2530,receive and buffer terminal data;and use the wifi function in wireless network and nodemcu module to realize real-time collection of field environmental data in agricultural greenhouse by MQTT communication protocol.Secondly,using the collected data and the meteorological observations published by China Weather Network,the prediction modeling of the core parameters and penalty factors of Support Vector Machine optimized by Grid Search at the highest and lowest temperatures in greenhouse is studied under the condition of considering different influence factors,and the time-to-time conversion coefficient is used to calculate the temperature at the corresponding time in greenhouse,so as to achieve the purpose of time-to-time prediction of temperature in greenhouse.In addition,under the full consideration of the influence of different environmental factors on greenhouse humidity,the hourly environmental humidity condition in agricultural greenhouse is predicted.In the study of greenhouse temperature early warning,using the algorithm of support vector classifier and decisiontree,the method of projection vector is used to calculate the separation degree between classes,to complete the construction of partial binary decision tree,to realize the early warning classification of temperature,and then to make the environment of agricultural greenhouse better suitable for crop growth.The experiment operating results show that the real-time and accurate environmental data reflected by this method provide a good theoretical support for the prediction and early warning of greenhouse temperature and humidity.This paper has carried on the practical application research to the agricultural greenhouse greenhouse temperature and humidity forecast and the early warning,uses the grid search optimized support vector machine to have the good modeling and the generalization ability to the small sample information data,and the forecast precision is high.This method promotes the development of real-time control of agricultural environment and provides theoretical basis and practical value to some extent.