摘要
为了解乌梁素海水质状况,提出一种基于M-GA-BP的乌梁素海悬浮物浓度反演方法,该方法以Sentinel-2遥感卫星影像为数据源,考虑研究区域存在的时空特性,提出将月份数据作为悬浮物浓度反演的一个特征,利用遗传算法(GA)优化BP神经网络的权值与阈值,建立GA-BP神经网络模型,并与传统BP神经网络模型进行比较。结果表明,引入月份特征模型有效降低了模型复杂度,提高了模型反演精度,其中,MGA-BP模型反演精度最高,训练集、测试集决定系数分别为0.916、0.903,训练集、测试集均方根误差分别为0.049、0.057μg/L,研究结果为乌梁素海悬浮物浓度反演提供了新思路。
In order to understand the water quality of Wuliangsu Lake,an inversion method of total suspended matter concentration based on M-GA-BP was proposed.Using Sentinel-2 remote sensing satellite images as the data source and considering the spatial and temporal characteristics existing in the study area,the monthly data was considered as a feature for the inversion of TSM concentration.The GA-BP model was built by optimizing the weights and thresholds of the BP neural network using genetic algorithm(GA),and comparing with the traditional BP neural network model.The results show that the introduction of the monthly feature model effectively reduces the model complexity and improves the model inversion accuracy,among which the M-GA-BP model has the highest inversion accuracy with the coefficients of determination of 0.916 and 0.903 for the training and test sets,respectively,and the root mean square errors of 0.049μg/L and 0.057μg/L for the training and test sets,respectively.The study can provide a new idea for the inversion of TSM concentration in the Wuliangsu Lake.
作者
吴陈昊
付学良
李宏慧
扈华
WU Chen-hao;FU Xue-liang;LI Hong-hui;HU Hua(College of Computer and Information Engineering,Inner Mongolia Agricultural University,Hohhot 010018,China)
出处
《水电能源科学》
北大核心
2023年第12期49-52,共4页
Water Resources and Power
基金
国家自然科学基金项目(62041211,61962047)
国家重点研发计划(2019YFC049205)
内蒙古自然科学基金(2020MS06011,2019MS06015)
内蒙古高校科技重点项目(NJZZ23044)。
