Trust Model Experimental Validation to Improve the Digital Twin Recommendation System

In the manufacturing domain, the digital twin has become an emerging concept for decision-making through the integration of what-if simulation capabilities. In such systems, the processing of the entire space of alternative solutions is very time-consuming; recommendation systems are used to solve this; however, these suffer from several problems, namely data sparsity and cold-start. The application of trust-based models can mitigate these problems, particularly the cold-start problems, by providing valuable background for the recommendation system. This paper presents the implementation and experimental validation of a trust-based model for improving the digital twin based what-if simulation recommendation system, addressing the cold-start problems. The proposed trust model was applied in an assembly line case study to recommend the best configurations for the optimal number of AGVs (Autonomous Guided Vehicles). The results show that applying the trust-based model with similarity metrics improved the mitigation of the cold-start problem.