Mobile Ad-hoc Networks (MANETs) are self-organizing, infrastructure-free networks that are vulnerable to Sybil and black-hole attacks. They are further hampered by energy limitations brought on by short node battery lives. In order to improve security, dependability, and energy efficiency, this study suggests an Enhanced Trusted Node Feedback Model (ETNFM) that combines energy-efficient clustering, machine learning-based malicious node identification, and trust-based feedback. The clustering, malicious node identification, and trust calculation processes are based on extensive mathematical derivations. ETNFM outperforms existing methods with great scalability for network sizes between 50 and 200 nodes, achieving a packet delivery ratio of 92.3%, 94.6% detection accuracy for malicious nodes, and 18% lower energy consumption, according to extensive simulations conducted in NS-3.