基于人工神经网络的沿海地区底泥盐度计算模型

Calculation Models for Sediment Salinity in Coastal Areas Based on Artificial Neural Networks

底泥盐度与海洋科学、河口研究、环境管理等密切相关,现有的底泥盐度计算公式存在精度不足、适用性有限等问题。为此,开展了271组室内试验和10组户外试验,整合了其他学者的研究数据,以底泥电导率、泥沙浓度、温度和细颗粒表面系数为模型输入变量,分别建立了用于计算沿海地区底泥盐度的反向传播人工神经网络(BP-ANN)模型、粒子群优化的反向传播人工神经网络(PSO-BP-ANN)模型、结合遗传算法的反向传播人工神经网络(GA-BP-ANN)模型。与现有的底泥盐度计算公式相比,新建模型的精度更高,可为沿海地区底泥盐度的确定提供更多可供选择的预测方法。

The salinity of sediment is closely related to marine science,estuarine research,and environmental management.The existing formulas for calculating sediment salinity have some problems,such as lack of accuracy and limited applicability.In view of this,this study carried out 271 sets of laboratory tests and 10 sets of field tests,and integrated the research data of other scholars.With sediment conductivity,sediment concentration,temperature and surface coefficient of fine particles as input variables,the back propagation artificial neural network(BP-ANN)model,particle swarm optimization back propagation artificial neural network(PSO-BP-ANN)model and genetic algorithm combined back propagation artificial neural network(GA-BP-ANN)model for calculating sediment salinity in coastal areas were established respectively.Compared with the existing sediment salinity calculation formulas,the new models have higher calculation accuracy and provide more alternative prediction methods for determining sediment salinity in coastal areas.