Design And Research Of Stage Precision Fertilization System For Forest Based On Android Platform

In forestry production,the calculation of the amount of N,P,K and other fertilizer is required in view of different forest species and different soil conditions.Fertilization prediction model,as the key to the implementation of precision fertilization in Chinese fir forest,has become a hot issue in forestry scientific research,but satisfactory accuracy and stability cannot always be obtained by adopting the traditional rationale fertilization model.As "Internet Plus" matures and prevails,it is an irresistible trend to develop intelligent forestry with the aid of "Internet Plus" in the future.Taking Chinese fir forest industry as an example,which is a dominant forestry industry of state-owned forest center in Renhua County,Guangdong Province and Liujiashan forestry center in Shixing County,this thesis determines the key factors affecting the fertilization rate of trees by adopting grey relational degree analysis.Besides,three kinds of neural networks are used to establish the prediction model for forest fertilization,and the factors with high grey relational degree are taken as input variables.The three prediction models are compared after verifying their validity,with the parameters such as site index,forest age and relevant nutrient content of dominant trees are taken as input of neural network,and the actual fertilizer application rate as the output.The results show that the fertilization model of GRA-PSO-BP neural network is superior to that of GRA-Elman neural network and GRA-LSTM Algorithm,it can effectively guide phased precision fertilization since its error between the predicted and actual fertilization amount is less than 5%.In view of this situation,this thesis chooses the GRA-PSO-BP neural network with high precision and good stability as the forest fertilization prediction model,and finally completes the design and research of the precision fertilization prediction system based on Android.The main findings are as follows:(1)The crucial influencing factor of forest fertilization amount is determined by applying the grey relational degree analysis,and then three different neural network models are established topredict forest fertilization,at last,the optimal GRA-PSO-BP neutral network prediction model for forest fertilization is selected through contrasting the three prediction models.(2)Based on the GRA-PSO-BP neural network forest fertilization prediction model,a forest fertilization system based on the Android platform is established.After analyzing the demand of the forest fertilization prediction system,the system is designed and studied according to the demand.The system not only realizes the function of forest fertilization prediction,but also includes related extended functions,which can meet the needs of forestry precision fertilization;(3)The systemic forest fertilization prediction and other corresponding functions are accomplished by applying Android technology.Compared with fertilization systems based on other platforms,the Android-based prediction system for forest precision fertilization is portible and convenient which can be further optimized as intelligent fertilization expert system for forests in future researches.