基于人工鱼群算法的冰情预报神经网络模型  

Neural Network Model for Ice Forecasting based on Artificial Fish Swarm Algorithm

作　　者：王富强[1] 魏怀斌[1] 

机构地区：[1]华北水利水电大学,河南郑州450011

出　　处：《人民黄河》2014年第1期30-32,共3页Yellow River

基　　金：国家自然科学基金资助项目(51009065);"十二五"国家科技支撑计划项目(2012BAC19B03);河南省重点科技攻关计划项目(112102110033)

摘　　要：在分析凌汛成因的基础上选取合适的预报因子,针对BP神经网络收敛速度慢、易陷入局部极小值的缺点,利用改进的人工鱼群算法训练BP神经网络,以黄河宁蒙河段封开河日期数据进行建模,给出了人工鱼群算法训练神经网络的基本原理和步骤,并对人工鱼群算法神经网络模型、遗传算法神经网络模型、粒子群神经网络模型的预测结果进行了对比分析。结果表明:人工鱼群算法神经网络模型对黄河内蒙古段凌汛期的封开河日期预测比较准确,预测结果优于遗传算法神经网络模型和粒子群神经网络模型。Based on the analysis on the factors affecting the formation of ice-jam flood,the most important factors for forecasting were selected. Ac- cording to weak points of slow convergence and being apt to local minimum about BP neural network, adopting artificial fish-swarm algorithm was suggested to train the artificial neural network. According to the freeze-up and break-up date of Ningxia-Inner Mongolia section of the Yellow River, neural networks had been trained by adopting AFSA to build an AFSA-NN ,which was realized by MATLAB 7.0 and employed to forecast ice flood. The case study shows that this algorithm forecast is correct for freeze-up and break-up date in the ice flood season in Inner Mongolia reach of the Yellow River. The forecast result is better than GA-BP and PSO-BP neural network.

关 键 词：神经网络 人工鱼群算法 封河日期 开河日期 宁蒙河段 黄河 

分 类 号：TV213.4[水利工程—水文学及水资源] TV882.1