Abstract: There are many hidden safety dangers and dynamic changes in the construction process of hydropower projects, which is the main reason for the frequent accidents during the construction of hydropower projects. In order to evaluate the degree of the potential safety dangers in the period of construction, an evaluation model for the hidden safety dangers of the hydropower construction is established based on fuzzy comprehensive evaluation and BP neural network. A multi-level with multi-index system for the diagnosis of the hidden dangers during the construction is developed, and five levels, namely major, large, general, small and slight levels, are proposed. Using the example, the hydropower construction is evaluated, the safety hazard level is determined, and the evaluation results are analyzed. The evaluation results of the potential safety dangers show that the hidden safety danger level of the example station is the third level, which conforms to the actual situation of hidden danger investigation. The fuzzy neural network model proposed in this paper has strong operability and can effectively analyze the potential hidden safety dangers during the construction. It has a certain reference value for the investigation of the potential hidden safety dangers in the construction process of hydropower projects.