Intelligent Harmonic Suppression in Multilevel Inverters Using RL-Enhanced NMPC for Power Quality Optimization

Power system performance requires the development of improved Total Harmonic Distortion (THD) reduction methods for multiple-level inverters (MLIs), which function in high-voltage industrial applications. NVTC serves as one of the main traditional approaches that have successfully suppressed harmonics in various applications. The phase voltage THD reduction effectiveness of NVTC remains satisfactory, but the method neglects dominant line-to-line voltage noise that affects system efficiency and load performance. The real-time adaptability of NVTC deteriorates when dynamic system changes and disturbances appear, which generates suboptimal performance during changing load conditions. The proposed optimization design merges NMPC with reinforcement learning to create a sophisticated method for achieving precise reduction of line-to-line voltage THD. Real-time system performance improves through the RL-NMPC framework by permitting an RL agent to continuously acquire ideal switching strategies for minimizing harmonic distortion and a hard real-time NMPC controller for operational stability maintenance. Concurrently operating these elements enables the system to perform 50% better regarding THD reduction along with a power quality boost of 25–30% beyond standard NVTC operation. The analysis demonstrates that power quality index (PQI) along with switching efficiency and electromagnetic interference (EMI) experiences 15–20% reduced switching losses together with 10–15% decreased EMI. The validated framework demonstrates real-time capabilities and industrial MLI system scalability based on MATLAB/Simulink simulation validations. The advanced RL-NMPC framework presents prospects for upcoming power converter systems because it matches industry demands for powerful distribution systems with high efficiency and reliability while remaining clean.