<p>This paper presents a novel NOMA-based communication framework enhanced by Reconfigurable Intelligent Surfaces (RIS) and powered through vibration-induced mechanical energy harvesting. The study focuses on an underlay cognitive radio network, where the secondary transmitter regulates its transmit power according to both the harvested mechanical energy and the interference temperature constraint at the primary destination. To maximize the throughput of the secondary system, we jointly optimize the energy harvesting duration and the adaptive transmit power allocation. Furthermore, the impact of RIS element density on overall system performance is investigated. Simulation results reveal that the integration of RIS and adaptive power control substantially improves throughput under diverse vibration intensities and interference thresholds. The proposed approach demonstrates that vibration-based energy harvesting combined with NOMA and RIS can achieve energy-efficient and spectrum-aware communication, offering a promising solution for low-power, spectrum-constrained wireless networks.</p>