说明：  本文基于遗传算法思想，采用浮点数矩阵表示编码，在遗传算法的进化过程中加入一定的约束条件等方法，探讨了网络结构的设计和学习。经实例分析，在用于建立大坝安全监控预报模型的前馈神经网络设计中，该方法在满足一定约束条件下，能同时有效地寻找合适的网络结构和相应的参数(神经网络的权值和阈值)，且在精度和速度上都有较大的提高，为实现实时在线分析评价大坝的安全性态提供了有力的技术支持。
(Based on the genetic algorithm, using a float matrix coding, Genetic algorithms in the evolutionary process to be bound by certain conditions, to explore the structure of the network design and learning. By analyzing the examples used in the establishment of dam safety monitoring forecasting model of neural network design, The constraint in meeting certain conditions, can effectively find suitable network structure and the corresponding parameters (the neural network weights and thresholds), and the accuracy and speed have improved greatly. To achieve real-time online analysis and evaluation of the safety of the dam states provide strong technical support.)