西北工程技术学报2025，Vol.24Issue(3)：257-261,5.
基于BP神经网络和粒子群优化算法的林木蒸腾耗水模型
Forest Water Consumption Model Based on BP Neural Network and Particle Swarm Optimization Algorithm
摘要
Abstract
In order to quantify and simulate forest water consumption through transpiration,a thermal diffusion stem flow meter was applied in conjunction with the simultaneous measurement of meteorological factors to analyze the characteristics of forest water consumption and its influencing factors.A back propagation(BP)neural network and particle swarm optimization(PSO)algorithm were employed to simulate the transpiration water consumption of Euonymus bungeanus in the eastern sandy area of the Yellow River.The results indicated that solar radiation,air temperature,the difference in saturated vapor pressure,and relative humidity are the main influencing factors of forest water consumption,which were selected as input variables for the model.The simulation accuracy of the forest water consumption models established on BP neural network and PSO-BP neural network was above 0.77.Compared to the BP neural network model,the PSO-BP neural network model improved the determination coefficient by 11.69%,the Nash efficiency coefficient by 17.81%,and reduced the average absolute error by 20.00%.It is evident that the PSO-BP neural network model better reflects the relationship between tree transpiration water consumption and meteorological factors,demonstrating improved learning efficiency,stability,and prediction accuracy,which is significant for the widespread application of neural network prediction models.关键词
林木/蒸腾耗水/BP神经网络/粒子群优化算法/宁夏Key words
trees/transpiration consumes water/BP neural network/particle swarm optimization(PSO)algorithm/Ningxia分类
农业科技引用本文复制引用
周鹏,马军,马云蕾,王娜娜,柳利利,韩磊..基于BP神经网络和粒子群优化算法的林木蒸腾耗水模型[J].西北工程技术学报,2025,24(3):257-261,5.基金项目
宁夏自然科学基金项目(2023AAC03056) （2023AAC03056）
