As modern vehicle architectures grow increasingly complex, the automotive industry relies on Model-Based Systems Engineering (MBSE) to design systems and ensure traceability. MBSE integrates requirements with functional, logical, and physical views for cohesive system design. This study investigates automating functional architecture generation from requirements using advanced clustering techniques. Although clustering has been widely applied to non-functional requirements, its use for functional requirements (FRs) remains underexplored. To address this, FRs were transformed into dense semantic vectors using embedding models such as BERT, ELMo, GloVe, and Doc2Vec. These embeddings enabled precise clustering through Agglomerative Hierarchical Clustering (AHC) to identify semantic relationships. Results show that BERT embeddings outperformed other methods, achieving the highest clustering accuracy and precision. This highlights the potential of large language models (LLMs) to advance the clustering and classification of FRs by enhancing contextual understanding and scalability. The study demonstrates that leveraging advanced NLP techniques significantly improves functional requirement management in automotive systems. This approach supports more effective decision-making, enhances design efficiency, and streamlines system development, advancing MBSE practices in handling complex automotive architectures.