基于流场原位实测的太湖水动力驱动关键模型参数研究

By means of a tracer device that simulates the movement of water particles and its real-time positioning, the in-situ measurement of the flow field in Taihu Lake is carried out to obtain the real-time spatio-temporal distribution of the flow field in Taihu Lake under typical wind conditions. According to the status of vegetation and sediment at the bottom of Lake Taihu, an experimental study on the bottom friction coefficient at Lake Taihu was conducted in the rectangular flume at Hohai University aims to obtain different bottom friction coefficients of different regions of Lake Taihu. Construct a gas-water coupling hydrodynamic model of Lake Taihu which considers the interaction of wind field, wave and lake flow. Based on the in-situ measured data of the flow of Lake Taihu, undertake the uncertainty and sensitivity analysis of the model parameters using Latin Hypercube Sampling (LHS) and standard rank stepwise regression, to obtain the universal Taihu hydrodynamic model parameter values suitable for different models and different grids, and further improve and develop the hydrodynamic mathematical model of Taihu Lake. Under the condition of typical wind farm, study the movement rules of water body mass in different areas of Taihu Lake and the change of water age in order to evaluate and explain algae accumulation and blooming in Taihu Lake.

通过一种可模拟水流质点运动并实时定位的示踪装置，开展太湖流场原位测量，获取典型风场条件下太湖流场实时时空分布状况。根据太湖湖底植被及底泥状况，在河海大学矩型水槽内开展太湖底摩擦系数的实验研究，得到太湖不同分区的湖底摩擦系数。构建考虑风场、波浪、湖流共同作用的太湖气-水耦合水动力数学模型；基于太湖湖流的原位实测资料，采用拉丁超立方取样法( LatinHypercube Sampling，LHS)以及标准秩逐步回归法开展模型参数的不确定性及敏感性分析，得到适用于不同模型、不同网格划分的普适性太湖水动力模型参数值，进一步完善和发展太湖水动力数学模型。开展典型风场条件下太湖不同区域水体质点在太湖的运移规律以及水龄变化研究，从而对太湖藻类堆积、水华暴发等现象做出评估及解释。

基于室内水槽实验，对不同槽底条件下的情况进行实验分析，发现在流速一定的情况下，曼宁系数与太湖水生植物密度呈正相关关系，太湖实际曼宁系数会随水生植物密度的增加而增大，全太湖不同湖区的曼宁系数变化范围在0.002~0.032s/m1/3，结果为太湖水动力模型的构建提供了关键的曼宁参数基础。2019年期间，课题组进行了两次野外原位流场实验，得到了太湖主要几个湖区的实时流速为西部沿岸区（湖面开阔区，16.01 cm/s）>梅梁湾（湖湾开阔区，14.40 cm/s）>贡湾（湖湾开阔区，12.25 cm/s）>竺山湾（湖湾狭窄区，10.7 cm/s）>东湖区（岛屿群聚区，9.39 cm/s）。结合构建的水动力数学模型率定出风拖曳系数分别为0.08和0.23，并对边界风场进行了泰森多边形计算，提升了水动力模拟精细化程度。基于近五年的基础数据，结合LHS抽样方法，借助已构建好的太湖水质机理模型，模拟得到182500组计算方案，根据精准度和稳定度评估筛选出VAE-LSTM作为水质替代模型，耦合GWO优化算法和K-mean聚类算法形成了达标方案制定方法体系。