基于PG-BP神经网络的流域洪涝预测及仿真  

Basin Flood Prediction and Simulation Based on PG-BP Neural Network

作　　者：韩卓慧 闫长青[2] HAN Zhuo-hui;YAN Chang-qing(College of Computer Science&Engineering,Shandong University of Science and Technology,Qingdao 266590,China;College of Intelligent Equipment,Shandong University of Science and Technology,Tai’an 271019,China)

机构地区：[1]山东科技大学计算机科学与工程学院,山东青岛266590 [2]山东科技大学智能装备学院,山东泰安271019

出　　处：《软件导刊》2021年第5期50-55,共6页Software Guide

基　　金：中国科学院先导专项(A类)子项目(XDA20030203);山东省高等学校科技计划项目(KJ2018BAN072);全国统计科学研究项目(2016LZ12)。

摘　　要：河流水位数据是洪涝灾害仿真模拟的重要依据,而精准的水位预测可以给洪水的淹没范围提供可靠的参考信息。单一BP神经网络的水位预测模型通常用于洪水水文模拟,但其准确度不高。为了精确分析洪水淹没范围并实现洪水淹没的仿真模拟,首先引入主成分分析方法(PCA)提取出影响水位变化的主成分,然后将主成分作为GA-BP神经网络的输入变量,河流水位数据为输出变量,建立PG-BP神经网络洪水水位预测模型。以大沽河流域为研究区域,使用该模型对汛期水位进行预测,根据模型预测的水位数据,可实现大沽河流域洪水淹没的仿真模拟。该模型水位预测的预报准确率均值达99.8%,预报效果较好,拟合精度较高,且可视化仿真也能够真实生动地显示出受灾地区,可以为防洪决策提供有力支撑。River water level data is an important basis for flood disaster simulation,and accurate water level prediction can provide reliable reference information for flood inundation range.Single BP neural network water level prediction model is usually used to simulate flood hydrology,but its accuracy is poor.In order to accurately analyze the flood submergence range and realize the simulation of flood submergence,the principal component analysis(PCA)method is introduced to extract the principal component affecting the water level change.Then,the principal component is used as the input variable of GA-BP neural network,and the river water level data is the output variable.A pg-bp neural network flood level prediction model is established.In this paper,the model is used to predict the water level of Dagu River in flood season.According to the water level data predicted by the model,the simulation of flood inundation in Dagu River Basin is realized.The average prediction accuracy of the model reaches 99.8%,the prediction effect is good and the fitting accuracy is high.The visual simulation can also show the disaster area vividly,which can provide strong support for flood control decision-making.

关 键 词：BP神经网络 水位预测 洪水淹没 仿真模拟 水文分析 

分 类 号：TP306[自动化与计算机技术—计算机系统结构]