基于改进粒子群算法的模糊神经网络及其在短时天气预报中的应用

AN IMPROVED PSO-BASED FUZZY NEURAL NETWORK AND ITS APPLICATION IN SHORT-TERM WEATHER FORECAST

提出一种基于改进粒子群模糊神经网络进行短时天气预测的方法,将粒子群算法与模糊人工神经网络进行融合,充分发挥粒子群算法全局寻优的优势。以上海地区天气预报作为实例,建立了基于改进粒子群算法的多模型模糊神经网络预报模型,试验结果表明该方法对于短时天气预报具有较好的准确度,得到了上海中心气象台有关专家的肯定。

This paper discusses an approach for practical short-term weather forecast based on improved PSO（Particle Swarm Optimization） fuzzy neural network,in which the fuzzy artificial neural network is fused with the particle swarm optimization to put the advantages of PSO＇s global best search into full play.Taking the weather forecast of Shanghai region as example,the improved PSO-based multi-model fuzzy neural network forecasting model is established.Experimental results show that the multi-model FNN（Fuzzy Neural Networks） for short-term weather forecasting has an acceptable accuracy,it has been approved by the experts of Shanghai Central Meteorological Observatory.