利用多个自组织神经网络的同时训练 ,可以一次性自动地生成计算区域内与计算区域边界上所有网格节点 ,尔后按 FGT法的基本思想将节点连接成所需的三角形或四边形单元。由于在自组织神经网络中采用了改善的目标函数 ,节点的分布可以实现自适应调节以反映网格疏密分布的要求。文末几个算例验收证明本算法具有自适应性 ,适用于凸域、凹域 ,多连通域等多种情况 With simultaneous training of multiple self-organizing neural networks, all grid nodes in the calculation area and the boundary of the calculation area can be generated automatically at one time, and then the nodes are connected into the required triangle or quadrilateral elements according to the basic idea of ​​FGT method. Due to the adoption of an improved objective function in the self-organizing neural network, the distribution of nodes can be adaptively adjusted to reflect the requirement of dense distribution of grids. Several examples at the end of the acceptance test shows that the algorithm is adaptive, suitable for convex domain, concave domain, multiple connected domains and other circumstances