A Study Of Tree Growth Model Based On BP Neural Network Of Multi-factor Fusion

Forest resources are the material basis for the development of human survival.They can not only provide a variety of material and raw materials for development and production,but also play an important role in the stability and healthy development of the entire ecological environment.Therefore,it is of great significance to accurately calculate the growth data of trees.At present,in the actual investigation,the data of DBH and tree height are generally measured and collected by traditional measuring instruments(such as tape,caliper,etc.),and then recorded by manual means.However,the traditional data collection methods have the following problems:(1)The variety of parameters that need to be collected often leads to misdetection or missed measurement;(2)There are many trees that need to be measured in forestry,and a huge workload requires a lot of time and energy.Therefore,the use of advanced technology for effective prediction of tree growth is of great significance to forest scientific management.BP artificial neural network has the advantages of self-learning,high-speed search for the optimal solution,etc.In recent years,BP neural network has been gradually applied in forestry.Aiming at the problems existing in the field of forestry data collection and statistics,and the shortcomings in the field of forest land assessment,in this paper,a BP neural network tree growth model based on multi-factor fusion was designed for the first-class national inventory of arbor plantations in Jingning County in 2004,2009 and 2014.Using Pearson correlation coefficient,several correlation factors with high correlation are selected as the prediction basis of tree growth model based on the correlation between correlation factors and factors to be predicted,and then the BP neural network structure with multiple inputs and three outputs is constructed.The BP neural network is optimized by genetic algorithm to improve the accuracy of BP neural network.Finally,the neural network is trained by using the selected data,which is a tree growth model.The purpose of this model is to predict the average tree height,average DBH and standing stock volume of trees by inputting several forest impact factors,and compared with other models.The results showed that the total relative error of DBH was 5.89%,the total error of average tree height was 2.77%,and the total error of standing stock was 3.74%.The total error of the model was lower than that of other models,which indicated that the growth model designed in thispaper had better prediction ability for stand growth and can meet the needs of practical applications.