基于深度置信神经网络模型的城区防洪预警模型的设计  

Design of Urban Flood Control Early Warning Model Based on Deep Confidence Neural Network Model

作　　者：蒋啸 JIANG Xiao(Jiangsu Mobile Information System Integration Co.,Ltd.Nanjing 210029,Jiangsu)

机构地区：[1]江苏移动信息系统集成有限公司,江苏南京210029

出　　处：《长江信息通信》2025年第2期19-21,共3页Changjiang Information & Communications

摘　　要：针对传统的水文、气象防洪预警模式已不适应当前复杂的城市建设、居住现状,文章应用深度置信神经网络,设计了防洪预警模型。首先阐述了预警模型的需求,依据需求提出了基于深度置神经网络模型的防洪预警模型,并根据防洪特征,提出了改进模型DBN-DNN,详细论述了模型的训练过程,最后通过实验,本模型与BP神经网络、LSTM、SVM、DBN模型在AUC指标上做了对比分析,经验证本文模型优化其他模型。In response to the fact that traditional hydrological and meteorological flood warning models are no longer suitable for the current complex urban construction and residential situation,this paper applies deep confidence neural networks to design a flood warning model.Firstly,the requirements of the early warning model were elaborated,and a flood warning model based on deep neural network model was proposed according to the requirements.Based on the flood control characteristics,an improved model DBN-DNN was proposed,and the training process of the model was discussed in detail.Finally,through experiments,this model was compared and analyzed with BP neural network,LSTM,SVM,and DBN models in terms of AUC index.After verification,this model optimized other models.

关 键 词：深置置信网络 深度学习 DBN-DNN 防洪预警 AUC 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]