Unmanned Aerial Vehicle (UAV) avionics systems present complex multi-objective optimization challenges involving reliability, response time, power consumption, and system weight. Traditional optimization approaches often struggle with the computational complexity and dynamic nature of these systems. This paper presents a novel hybrid Genetic Algorithm-Particle Swarm Optimization (GA-PSO) algorithm enhanced with deep learning for real-time UAV avionics optimization. Our approach integrates adaptive weight adjustment mechanisms, deep learning-enhanced fitness evaluation, and comprehensive multi-objective optimization frameworks. Experimental results demonstrate significant improvements over baseline methods, with 40–60% reduction in computational overhead, 2–4 times faster convergence, and 15.7%, 32.4%, and 18.2% improvements in reliability, response time, and power consumption efficiency, respectively. The system supports real-time reconfiguration within 5 s for dynamic obstacle avoidance, making it suitable for practical UAV applications.