In recent years, unmanned aerial vehicles (UAVs) have seen significant growth in both military and civilian applications. Among them, rotary-wing UAVs (RUAVs), particularly multi-rotors, have gained attention from researchers for designing autonomous control systems due to their vertical takeoff and landing capabilities and rapid maneuvers. Most traditional multirotor control methods are based on First Principles Techniques (FPT), which face challenges in managing uncertainties. Recently, intelligent fuzzy systems have replaced these methods; however, despite improved performance, their fixed structure limits adaptability to dynamic environments. Evolving fuzzy systems (EFS), by updating both structure and parameters, provide an effective solution for managing sudden changes in real-time multirotor flights. Although advanced EFS versions have been developed, they often focus on system identification, neglecting sufficient parameter optimization. This research introduces a new EFS called GODFIS (Globally Optimized Dynamic Fuzzy Inference System), which utilizes recursive center-of-gravity and global optimization of consequent parameters to address this gap. Additionally, a new criterion, the "Noise Elimination Principle" (NEP), is proposed to maintain response quality and reduce noise effects. Experiments demonstrate that GODFIS achieves higher accuracy than existing methods in benchmark problems. Using diverse datasets, it has been proven that this algorithm not only excels in multi-rotor modeling and control but also offers high generalizability and broad applicability in various control and prediction domains.