A Predictive Autorotation Entry Analysis Using Bayesian Multi-Model Estimation Detection

Abstract

Improvements to achieving helicopter autorotation in an event of engine or driveshaft failure is an issue that has and always will be present in the engineering landscape. Questions include, “how much of a performance hit can be taken for the proper autorotation tools?” and “how much control should the pilot initially have in the event of an engine failure?” One certain aspect is that upon an engine failure in a rotorcraft, every second counts. A prediction of future behavior of a rotorcraft can be estimated by feeding observational flight data into comparable dynamic and engine analysis models. The comparison of different hypothesis models is combined into a Bayes’ Multi-Model Estimation to evaluate the health of the rotorcraft. Interesting concepts through this work are (a) the creation of a coupled dynamic and engine model that is iterated into failure models for detecting a risk, (b) the methodology for using multiple models for reducing false positives in engine failure detection, and (c) the magnitude of the change in pilot recognition time. The prediction of a failure in the rotorcraft model can lead to new contributions of methods into rotorcraft autorotation detection. The use of live observation data to project the state and health of the rotorcraft at a future time has been shown to be valid. The failure models were to be generated by reasonably changing states and controls in the normal functioning model. As predicted, the aircraft altitude saved due to the significant reduction in pilot recognition time is essential for increasing the success of an autorotation landing.