Abstract:

To address the problems of delayed state perception, single monitoring dimension, and insufficient visualization in the quality inspection equipment for nuclear power connection sleeves, a state monitoring method driven by digital twin was proposed. A digital twin-based collaborative state monitoring framework for the inspection equipment was constructed. Based on the OPC UA technology, a multi-source information interconnection model was established. A finite state machine model was employed to discretize and logically drive the inspection process, and a hierarchical verification strategy was proposed to establish a multi-dimensional motion state monitoring mechanism. A surrogate model coupling the radial basis interpolation function and Gaussian process regression was constructed. The experimental results indicate that the proposed method achieves an average motion state anomaly detection accuracy of 98% and an alarm response time of less than 0.5 s. While preserving the fidelity of mechanical data, the computation time is reduced to 0.15 s, achieving an R² of 0.92 for stress prediction and 0.81 for strain prediction, effectively balancing the accuracy and efficiency of mechanical state monitoring under complex working conditions. This method provides an effective solution to meet the "full inspection and full qualification" requirements for nuclear power containment rebar connection sleeves.

Key words: digital twin, surrogate model, finite state machine model, state monitoring, nuclear power connection sleeve quality inspection equipment