Asset Administration Shell (AAS) and its submodels are essential for achieving standardization and interoperability in Industry 4.0 environments. To ensure standardization in industrial settings, our previous research focused on automatically validating standardized submodels according to the Industrial Digital Twin Association (IDTA) standards and non-standardized submodels based on user-defined templates. Although earlier work focused on automatic submodel validation triggered via MQTT, the semantic correctness of the elements remains a critical, yet underexplored area. This paper extends the Test Orchestrator framework that leverages a semantic dictionary such as ECLASS and Generative AI for increased semantic validation capabilities. These enhancements enable validation of element naming, unit appropriateness, and compatibility of values with declared value types, thereby improving the contextual accuracy of submodels. Additionally, the system introduces feedback generation for incorrect or inconsistent elements, helping users better understand and resolve validation issues. A large-scale scalability evaluation demonstrates the framework’s ability to efficiently handle a significant number of submodels while maintaining minimal per-submodel validation time. The proposed solution contributes to more robust, intelligent, and semantically aware digital twin systems that are more closely aligned with the objectives of Industry 4.0.