河网汊点分组解法的自动分组技术及优化

Automation and optimization of junctionsgroup method in river network

针对河网计算中应用较广的汊点分组解法,分析其无法实现分组自动化的不足之处.并在汊点分组解法的基础上,将矩阵计算中常用压缩矩阵带宽的Cuthill-Mckee法进行了变换,使其能够实现对汊点进行的排序,然后用拆组法对汊点进行分组,最终形成了无需人工干预的自动分组技术.分组后,还根据其系数矩阵的特点,结合Cuthill-Mckee法和保带状结构三角分解的Doolittle分解,形成稀疏对称存在矩阵求逆算法,在河网系数矩阵求逆中取得了良好的计算效果.最后以长江中游洞庭湖区的河网计算为例,对比原方案与改进后的计算耗时,取得了令人满意的结果.说明了该改进能有效地提高汊点分组的计算效率.

The shortcomings of the junctions group method are analyzed; and the Cuthill-Mckee method for numerical simulation in river networks is applied on the basis of the junctions group method, which is called automated junctions group method. The authors still compressive grouping matrix by integrating Cuthill-Mckee method into Doolittle method; and good results are obtained. Finally, the Dongting Lake model after optimized is compared with the original one; and the satisfactory result proves that it can effectively save the computational time.