In unmanned aerial vehicles (UAVs), an inertial measurement unit (IMU) sensor is essential for maintaining stability and navigational accuracy during flight. It becomes exceptionally crucial when UAVs undertake complex tasks, such as flying near wind turbines for inspections or maintaining precise formations alongside other UAVs. The main challenge stems from the nonlinear nature of IMU sensor readings, especially in situations requiring meticulous control. In this article, the authors suggest integrating an unscented Kalman filter (UKF) with a support vector machine (SVM) to predict defects for effective fault prediction in UAVs. The efficacy of this method is validated through comparative experiments with standard prediction algorithms, demonstrating its accuracy in various simulated faulty scenarios. As a result of this research, the proposed method can serve as a trend predictor for monitoring IMU failures as well as a method of enhancing the reliability of drones.