Compared to traditional core damage assessment systems, in this paper, artificial intelligence technologies such as expert knowledge bases and neural network models are innovatively introduced and a new system development approach is proposed. Firstly, an overview of the construction of expert knowledge base and the training methods of deep neural network models are described. Based on RNN technology and the use of nuclear accident data, the network model was trained to establish a function mapping between input data and labels, which enable the model to accurately calculate the core state as much as possible. Then, the trained network model is applied to the development of a nuclear power plant core damage assessment system. The framework of system, key technologies, and application are introduced for the development of new core damage assessment system. It not only compensates for the shortcomings of traditional CDAG calculation methods that are rough and not intuitive, but also is an important exploration of the application of artificial intelligence algorithms in the nuclear industry. Which will effectively promote the digital and intelligent transformation of nuclear power plants.