改进遗传神经网络方法在大气环境质量评价中的应用

Improved Genetic Neural Network and Its Application in Atmospheric Quality Assessment

运用双极性压缩函数适应度定标和基于排挤方法的选择算子改进标准遗传算法（ＳＧＡ），使其成为简单通用、快速收敛的并行全局搜索算法。利用该算法优化误差反向传播网络（ＢＰＮ），克服了ＢＰＮ收敛慢和不具有全局收敛性的缺陷，在此基础上，建立大气环境质量评价模型，并将该模型用于实例。结果表明，该方法用于大气环境质量评价是可行的。

This paper presents a method that improves the simple genetic algorithms (SGA) with bipolar compressed functional fitness scaling and selection operator based on crowding model.The improved genetic algorithm(GA) becomes a kind of general-purpose global search algorithm.We use the method to optimize back-propagation neural network(BPN),and put forward a model that evaluates the grade of air pollution based on it .The experimental results are also presented to show that the method is feasible.-