大坝的结构和地质条件复杂，荷载种类众多，常规专家系统采用的知识表示方法和推理机制有时难以及时准确地对大坝的工作性态作出评价。本系统提出了传统专家系统与神经网络相结合的原理及实现方法，详细论述了系统的知识表示、学习、推理的机理和功能。系统已成功地应用于某混凝土面板支墩坝的疑点物理成因分析。 Due to the complex structure and geological conditions of the dam, numerous types of loads and the knowledge representation and reasoning mechanism adopted by the conventional expert system, it is sometimes difficult to evaluate the dam’s working behavior timely and accurately. The system puts forward the principle and method of combining traditional expert system and neural network, and discusses the mechanism and function of system knowledge representation, learning and reasoning in detail. The system has been successfully applied to the doubtful point of a concrete face pier.