Advanced DQN-MSOA based control architecture for power quality enhancement in multi-input renewable integration

Purpose This paper aims to present an advanced hybrid renewable energy system integrating photovoltaic (PV), wind and fuel cell sources with a 15-level modular multilevel inverter (MMI) to enhance power quality and system efficiency. The aim is to improve energy utilization, stability and voltage regulation, while minimizing harmonics, using advanced optimization techniques for renewable integration. Design/methodology/approach A novel hybrid Deep Q-Network (DQN) and Modified Sunflower Optimization Algorithm (MSOA) is proposed for real-time control and optimal parameter tuning. The DQN provides adaptive decision-making under dynamic operating conditions, while MSOA optimizes switching angles and controller parameters to minimize harmonic distortion and improve system stability. In addition, a hybrid proportional integral derivative (PID)–PID–fractional order proportional integral derivative (FOPID) controller is used to enhance dynamic response and voltage regulation. Findings Simulation and experimental results demonstrate the effectiveness of the proposed approach, achieving a Total Harmonic Distortion (THD) of 4.68% in simulation and 4.53% in hardware, both in compliance with institute of electrical and electronics engineers (IEEE) standards. The results confirm improved power quality, faster convergence and enhanced reliability compared to conventional methods, making the proposed framework suitable for next-generation hybrid renewable energy systems. Originality/value The originality lies in integrating of PV, wind and fuel cells with a 15-level MMI and an artificial intelligence-based control strategy. The use of a hybrid DQN-MSOA for power flow optimization and controller tuning is a novel approach to renewable energy integration.