Standards-Aligned AI Validation and Certification Platform for Trustworthy Modeling
Abstract
Artificial intelligence has been progressively implemented in engineering fields. However, a systematic framework for the validation, explanation, and certification of AI models was still lacking. We developed AIVeritas as a standards-aligned platform that operationalises the requirements of ISO/IEC 24028, 23053, 25023, 4213, and 17065 for the entire AI lifecycle in this research. The platform had integrated four components—data-quality analysis, model-performance evaluation, explainability diagnostics, and governance-based certification—to provide measurable, auditable, and reproducible assessments of AI systems. Experimental results had shown that each module had been indispensable to reliability, traceability, transparency, and certification readiness; the elimination of any module had always led to a decrease in trustworthiness scores. In addition, the platform had enabled rigorous verification in various engineering contexts through clause-level standards mapping, trust-index computation, lifecycle traceability, and explanation-stability analysis. The findings confirmed that AIVeritas had offered a unified and regulation-ready pathway for assessing the validity, robustness, and explainability of AI models used in engineering design and operational decision-making.