<p>In this work, a reliability simulation algorithm is proposed to assess the reliability of the interface considering various uncertain parameters. First, the uncertain shear strength of the interface is investigated through experiments and a Bayesian probabilistic approach. Then, an efficient surrogate model is constructed to predict the shear stress response using the deep learning method and simulation results. By involving uncertain material parameters, the time-varying shear stress response is efficiently predicted. Ultimately, the stress-strength interference model is used to assess the reliability of the interface at varying times and different locations. The results indicate that the proposed surrogate-assisted reliability assessment method could evaluate the interface strength of embedded sensors efficiently and accurately. The reliability of the interface is contingent upon the time-varying external load and the sensor locations within TPS.</p>