摘要
为提高神经网络预报模型的泛化能力和预测精度,将粒子群算法与BP神经网络相结合,建立了粒子与神经网络参数的映射关系,利用粒子群算法全局搜索网络的最优权值,赋值于神经网络进行训练,从而建立了PSO-BP预报模型,并将模型应用于黄河宁蒙河段的冰凌预报中。结果表明,该模型预报合格率较高,属于甲等预报,且与GA-BP模型相比偏差更小、精度更高。
In order to improve both the generalization capability and accuracy of neural network prediction model, particle swarm optimization （PSO） and BP neural networks are combined. Though building the mapping relationship between the particles and neural network parameter, the optimal weights obtained by PSO＇s global search are assigned to the neural network for training. So, the PSO-BP forecast model is established. This model is applied to Ningxia-Inner Mongolia reach of the Yellow River. The results which meets the requirement of class A of predicting level show that the proposed model has ; compared with the GA-BP model PSO-BP model is smaller and the accuracy is improved significantly e forecasting at the higher qualification the deviation of the
出处
《水电能源科学》
北大核心
2012年第3期35-37,172,共4页
Water Resources and Power
基金
国家自然科学基金资助项目(51009065)
河南省重点科技攻关计划基金资助项目(112102110033
092102110032)
河南省教育厅自然科学研究计划基金资助项目(2011B170006)
关键词
粒子群算法
神经网络
冰凌预报
封河
开河
particle swarm optimization
neural network
ice forecasting
freeze-up
break-up
