摘要
自从长短期记忆网络提出后,一般循环神经网络中存在的长期依赖问题得以解决,由于其独特的设计结构和良好的特性,适合对时间序列类型数据进行处理和预测,被广泛应用在机器学习和人工智能的各个领域中。针对锡林郭勒草原上的土壤湿度预测问题,选用岭回归算法、支持向量机算法、梯度提升决策树算法、长短期记忆网络算法和基于贝叶斯优化的长短期记忆网络算法进行对比实验,比较各个模型在回归分析中的平均绝对误差、均方根误差和平均绝对百分比指标情况。通过数据预处理和回归分析后,将长短期记忆网络算法应用在未来锡林郭勒草原上的土壤湿度预测,对自然环境保护和抑制草原沙漠化问题提供更多解决思路。
Since the short long-term memory network was proposed,the problem of long-term dependence in general recurrent neural networks has been solved.Because of its unique design structure and good characteristics,it is suitable for processing and forecasting time series type data,and is widely used in various fields of machine learning and human intelligence.Aiming at the prediction of soil moisture on Xilingol grassland,ridge regression algorithm,support vector machine algorithm,gradient lifting decision tree algorithm,short-term memory network algorithm and short-term memory network algorithm based on Bayesian optimization were selected for comparative experiments to compare the average absolute error,root mean square error and average absolute percentage index of each model in regression analysis.After data preprocessing and regression analysis,the long-term and short-term memory network algorithm was applied to predict the soil moisture on the Xilingol grassland in the future,providing more solutions to the problems of natural environment protection and grassland desertification control.
作者
肖天云
张子晨
魏佳妹
刘凤春
韩阳
XIAO Tian-yun;ZHANG Zi-chen;WEI Jia-mei;LIU Feng-chun;HAN Yang(The Key Laboratory of Engineering Computing of Tangshan,North China University of Science and Technology,Tangshan Hebei 063210,China;Department of Discipline Construction,North China University of Science and Technology,Tangshan Hebei 063210,China;COllge of Science,North China University of Science and Technology,Tangshan Hebei 063210,China)
出处
《华北理工大学学报（自然科学版）》
CAS
2023年第3期65-73,共9页
Journal of North China University of Science and Technology：Natural Science Edition
基金
国家自然科学基金面上项目(52074126)。
关键词
长短期记忆网络
贝叶斯优化
回归分析
土壤湿度预测
long short-term memory
bayesian optimization
regression analysis
soil moisture prediction
