To address the tracking control problem of tri-rotor unmanned aerial vehicles (UAVs) subject to simultaneous external disturbances and actuator faults, a predefined-time fault-tolerant tracking control strategy based on an iterative learning observer is proposed. Considering unknown time-varying disturbances and various types of actuator faults, a faulty dynamic model of the tri-rotor UAV is established. An iterative learning observer is designed to provide real-time estimation of disturbances and actuator bias faults, achieving asymptotic convergence of the observation errors without relying on an accurate system model. An adaptive algorithm is further introduced to estimate actuator loss-of-effectiveness faults, and an adaptive fault-tolerant controller with predefined-time performance is developed accordingly. The Lyapunov stability theory is applied to rigorously prove the convergence of the closed-loop system within the predefined time. Real-time experiments conducted on a flight control hardware-in-the-loop simulation platform demonstrate that, in comparison with finite-time sliding mode control, the proposed strategy not only guarantees system stability but also significantly enhances convergence speed and tracking accuracy, enabling precise tracking of the UAV within the predefined time.