A Novel Kalman Filter-Based Prognostics Framework for Performance Degradation of Quadcopter Motors

In the quadcopter, the performance of driving motor deteriorates with repeated flying cycles. However, the degradation is compensated by feedback control to maintain intended flight mission, which renders the state of degradation difficult to estimate until the failure occurrence. In order to solve this problem, a novel Kalman filter (KF)-based prognostics framework is proposed, which predicts the current health and remaining useful life (RUL) of the driving motors by taking advantage of the flight control data acquired during the flight, not by installing additional sensors. The framework consists of the online health diagnosis by the two-stage KFs to estimate the motor speeds and the maximum thrusts and the offline failure prognosis to predict the RUL of each motor by the particle filter algorithm. A Parrot Mambo drone is chosen to demonstrate the approach, in which the flight control data are collected during the hovering motion at regular intervals, while one of the motors undergoes accelerated degradation. The result shows that the framework predicts the RUL of the degraded motor with close accuracy against the failure threshold, which is the minimum thrust for takeoff of the quadcopter.