An enhanced fault diagnosis in nuclear power plants for a digital twin framework

Nuclear power plants can provide a huge amount of clean energy, which can help most countries to meet their greenhouse gas emission requirements according to the Paris agreement on climate change. To meet this energy need, the nuclear plant must be operated safely and economically, which makes the digital twin concept viable for achieving this aim. The digital twin can be used to monitor plant condition, fault diagnosis, prediction, and plant maintenance support systems. In this work, the framework for digital twin in a nuclear plant is proposed. This framework combines the application of the nuclear plant simulator and machine learning tools. The machine learning aspect of this digital twin concept is the focus of this paper. Data was generated by using a personal computer-based nuclear plant simulator. Principal component analysis was used in reducing the data dimension. Artificial neural networks and adaptive neuro-fuzzy inference systems were trained with the reduced data and used to diagnose the faults. Four faults in the plant were diagnosed with minimal error. The fault diagnosis is a significant aspect of the digital twin framework.