Unified Statistical Framework for Rotor Fault Diagnosis on a Hexacopter via Functionally Pooled Stochastic Models

In this work, a statistical time series method that is capable of effective multicopter rotor fault detection, identification, and quantification within a unified stochastic framework is introduced. The proposed framework is based on the functional model based method for fault magnitude estimation tackled within the context of statistical time series approaches. Estimator uncertainties are taken into account, and confidence intervals are provided for the fault magnitude of multicopter rotors. The framework employs functionally pooled (FP) models which are characterized by parameters that depend on the fault magnitude, as well as on proper statistical estimation and decision-making schemes. The validation and assessment is assessed via a proof-of-concept application to a hexacopter flying forward with a constant velocity under turbulence. The fault scenarios considered consist of the front and side rotor degradation ranging from healthy to complete failure with 20% fault increments. The method is shown to achieve fast fault detection, accurate identification, and precise magnitude estimation based on even a single measured signal obtained from aircraft sensors during flight. Furthermore, fault quantification is addressed via the use of both local ( boom acceleration) and global (IMU) sensors, with the signals collected from the boom supporting the identified faulty rotor proven to achieve better performance than the global signals, yet with a shorter signal length.