人工神经网络西太平洋副热带高压预报业务系统  

A WEST PACIFIC SUBTROPICAL PRESSURE OPERATIONAL PREDICTION SYSTEM OF ARTIFICIAL NEURON NETWORK

作　　者：冯民学[1] 陈必云[1] 袁瑞琴 韩桂荣[1] 周曾奎[1] 

机构地区：[1]江苏省气象台,210008

出　　处：《气象科学》1998年第4期396-402,共7页Journal of the Meteorological Sciences

摘　　要：本文应用常规500hpa资料，结合具有较稠密、高分辨的“TOVS”资料（卫星信息资料），分析、归纳了西太平洋副热带高压大幅度北移、南蓝（120°E脊线）的经验规则，在此基础上选择了具有最优下降因子的BP算法。在人工神经网络的结构方面设计了神经网络的前处理功能，将众多的单因子经基于专家系统的前处理模块，归纳的规则，作为网络的输入端，经网络学习、训练、迭代获得了较快的收敛速度和较高的历史拟合率，建立了“人工神经网络西太平洋副热带高压预报业务系统”In this paper,based on the combination using of routine 500hpa and Tovs da-ta,the experience rule of the large range moving of west Pacific subtropical high pressure wasconcluded,A Bp calculation method which has fine dropping factor was designed. In respectof network structure a pre - processing function was designed to be an inputting port.Through lots of studying and training,the faster convergence speed and higher history imi-tating have been got - After all,an artificial neuron network operational system was set up topredict the moving of subtropical high pressure.

关 键 词：人工智能 副热带高压 气压预报 业务系统 

分 类 号：P457.4[天文地球—大气科学及气象学] P732.7