| With the rapid development of forestry,people are increasingly demanding the quality of wood,the application scope of woods are becoming wider and wider,the cultivation of plantation has become one of the main points of concern.Climatic factors are an important factor which affect the growth of plantations.It is very important for the cultivation and operation of plantation to grasp the influence of climatic factors on the characteristics of wood correctly,which can make the forest resources be used rationally and effectively,and lay the theoretical foundation for the future research and development of forestry science.In this paper,the influence of climatic factors on the characteristics of wood and the Pinus koraiensis plantation was studied.(1)Using the method of multiple regression analysis to select the more evident characteristics of the wood affected by the climatic factors,according to the characteristics of the wood by the climate factor response to the moon,select the selected wood characteristics of the larger climate factors.(2)Based on the traditional RBF neural network(RBF),the prediction model of the influence of climatic factors on wood properties was established,and find out the shortcomings of traditional RBF neural network prediction.(3)By adding adaptive factors,an adaptive RBF neural network is proposed to improve the prediction accuracy and convergence speed of traditional RBF neural networks.(4)By using wavelet transform and BP neural network and RBF neural network,an adaptive wavelet RBF neural network is obtained in order to achieve higher convergence speed and prediction accuracy.Based on the results of the above process:(1)Through the form of the degree of influence on the characteristics of wood by climate factors.The length of late wood trachea in the growth rate and anatomical characteristics of the chemical characteristics of Pinus koraiensis plantation was selected as the wood characteristics selected in this study.Make the principal component analysis of the moon table,and make the correlation coefficient curve.Take the temperature in September,October,precipitation in November,relative humidity in December,sunshine time in September,October,the lowest ground temperature in July,the highest climatic factors in February as the eight climatic factors that affect the growth rate of Pinus koraiensis.The climatic factors such as the temperature in July,the relative humidity in October,the average temperature in July,and the lowest ground temperature in August were used as climatic factors for climatic factors affecting the tracheid length of Pinus koraiensis.(2)The prediction model was established based on the traditional RBF neural network by using the climatic factors of the corresponding characteristics as input and the growth rate of the wood and the anatomical characteristics as output..Using MATLAB for simulation,the simulation results show that the convergence rates of the growth rate and the growth curve of late tracheid length are 42 steps and 54 steps,respectively,and the average errors are 17.11% and 19.37% respectively.The convergence curve of the fitting curve is slower and the average error is larger.(3)The adaptive RBF neural network is improved by RBF neural network.The simulation results show that the convergence speed of the fitting curve is 39 steps and 48 steps,respectively,and the average error is 7.64% and 8.10% respectively.Error accuracy improved significantly,but the simulation speed to enhance less.(4)A hybrid algorithm based on wavelet transform and neural network-adaptive wavelet RBF neural network algorithm is proposed to further optimize the prediction accuracy and convergence steps of adaptive RBF neural network.The simulation results show that the convergence rates are 26 steps and 25 steps,respectively.The mean errors were 2.96% and 2.60%,respectively.Adaptive RBF neural network has a significant improvement in both the convergence speed and the average error of the adaptive RBF neural network. |
PDF Full Text Request