遗传神经网络在水质监测点优化布局中的应用

APPLICATION OF GENETIC NEURAL NETWORK IN LAYOUT OPTIMIZATION OF WATER QUALITY MONITORING POINTS

为科学全面地掌握湖泊水域的水质指标分布状况,以广西大学镜湖为实验研究对象,以无人船为平台将RBF神经网络与遗传算法相结合,对监测点的优化选择进行了讨论研究。分别对水质参数的单目标与多目标进行优化实验,并绘制拟合分布图。结果表明:遗传神经网络比传统的等距监测在水质参数的分布误差上具有明显优势,并在优化后至少6周内,其结果反映的数据依旧准确有效,说明遗传神经网络可明显提高水质监测的效率。

In order to grasp the distribution of water quality indexes in lake water scientifically and comprehensively, this paper took Jinghu Lake of Guangxi University as the experimental object, and unmanned ship as the platform, discussed the optimization selection of monitoring points based on combining RBF neural network with genetic algorithm. The single-objective or multi-objective optimization of water quality parameters was tested respectively, and fitting distribution map was formulated. The results showed that the genetic neural network had more obvious advantages in the distribution error of water quality parameters than the traditional isometric monitoring, and the data reflected by the results was still accurate and effective at least 6 weeks after optimization, indicating that genetic neural network could significantly improve the efficiency of water quality monitoring.