基于AGA-RBF神经网络模型的叶绿素a质量浓度预测研究  

The Study of Chlorophyll-a Mass Concentration Prediction Based on AGA-RBF Neural Network Model

作　　者：刘星宇 程建 牛艺晓[1,2] 杨春 LIU Xingyu;CHENG Jian;NIU Yixiao;YANG Chun(School of Mathematical Sciences,Sichuan Normal University,Chengdu 610066,Sichuan;Key Laboratory of the Evaluation and Monitoring of Southwest Land Resources,Chengdu 610066,Sichuan;Center of Network and Information,Sichuan Normal University,Chengdu 610066,Sichuan)

机构地区：[1]四川师范大学数学科学学院,四川成都610066 [2]四川师范大学西南土地资源评价与监测教育部重点实验室,四川成都610066 [3]四川师范大学网络与信息中心,四川成都610066

出　　处：《四川师范大学学报（自然科学版）》2024年第5期670-675,共6页Journal of Sichuan Normal University（Natural Science）

基　　金：国家自然科学基金(12101438);中央引导地方科技发展项目(2022ZYD0011);四川省自然科学基金(2022NSFSC1852)。

摘　　要：叶绿素a质量浓度是预测湖泊水华形成的重要影响因子,但常用的径向基(radial basis function,RBF)神经网络存在容易陷入局部极值,导致预测精度欠佳.针对这一问题,采用自适应遗传算法(adaptive genetic algorithm,AGA)对RBF神经网络进行优化,构建基于AGA-RBF神经网络预测模型,以莆田东圳水库为应用案例,对叶绿素a质量浓度进行预测,通过采集到的数据对预测模型进行仿真,对比均方根误差(RMSE)、相对误差(RE)以及平均相对误差(MRE),验证改进后的AGA-RBF模型具有更好的预测精度,以期对叶绿素a质量浓度进行长期预测.Chlorophyll-a mass concentration plays a crucial role in predicting the formation of lake blooms.However,the traditional radial basis function(RBF)neural network is susceptible to local optimal solutions,resulting in poor prediction accuracy.In this study,we employ an adaptive genetic algorithm(AGA)to optimize the neural network model,and a prediction model based on AGA-RBF neural network is constructed to predict the concentration of the chlorophyll a using Putian Dongshen Reservoir as an application case.Through the utilization of collected data to simulate the prediction model,we demonstrate that the improved AGA-RBF model has a good prediction accuracy,which is verified by the comparison of the root-mean-squared error(RMSE),the relative error(δ),and the average relative error(MRE).After the comparison of the root mean square error and the average relative error,it was verified that the improved AGA-RBF model has better prediction accuracy and is highly practical in the medium-and long-term prediction of the mass concentration for chlorophyll a.

关 键 词：RBF人工神经网络 自适应遗传算法 预测模型 叶绿素a质量浓度 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]