Development of health-monitoring algorithms for the diagnosis of motor phase faults in the electric propulsion system of a lightweight Remotely-Piloted Aircraft

Electrically propelled vehicles are growing trend, offering an answer to environmental requirements thanks to the reduction of CO2 emissions and an increasing energetic efficiency . Promising results in the development of new technologies oriented to an “electrification” of modern aircrafts has led to a great academic and industrial interest, bringing to life new applications which, in return, can provide useful new data that is gathered to improve the technology readiness level of these new approaches. The case study of this thesis is the conversion of the propulsion system of a lightweight remotely piloted aircraft (namely TERSA project) from an internal combustion engine to a fully electric powertrain, with a particular focus on the reliability analysis and the health monitoring of the architecture chosen to replace the old engine design. The first chapter describes the state of the art of powertrain systems used by Remotely Piloted Vehicles (RPAs) through an overview of items composing the propulsion system. Moreover, some techniques used for reliability assessment are shown alongside a case study used as reference for the next chapter. Chapter two starts with the definition of the reference mission that the aircraft is supposed to carry out in order to define the functional requirements that the items composing the powertrain system are summoned to perform. Thus, it is described the evolution of the architecture of the propulsion system applying iteratively various techniques aimed to demonstrate an improvement of system’s reliability until the final design. These techniques include Functional Hazard Assessment (FHA), Failure-Mode and Effects Analysis (FMEA) and Fault-Tree Analysis (FTA) and represent the standard suggested by international aeronautical regulations to demonstrate the safety of a system. Also, through the study of the behaviour of different components, possible faults that can occur in each of them are highlighted and data about their failure rate is gathered thanks to MIL-HDBK-217F and Ansi-Vita 51.1 standards. This passage is useful to stress out attention about critical items that need a redundant solution for a general improvement keeping low the cost and the weight of components. In fact, the TERSA design, which is adopted as the reference architecture in the following section of the work, employs two electric motors working in parallel capable of a complete decoupling, thus providing a great performance in terms of reliability. In chapter three is shown the evolution of the model representing the propulsion system through growing levels of complexity using MATLAB code and Simulink models. First, a DQZ model is used to study a relation between the input current provided to the system and the torque developed. A simple mathematical model of the aerodynamic torque produced by a propeller is used to subsequently develop a closed loop architecture to track the requested angular speed signal. Thus, a more accurate representation of the physics on which the system rely is carried out with a three-phase model answering the need to identify possible faults and correctly observe their effects on the rest of the system. The objective of this chapter in fact is to gather information about the time behaviour of systems dynamic parameters, stressing out the attention on how sensors can affect the performance of the control loop. Finally, this chapter displays the time response of the final TERSA architecture during normal operation. The fourth and last chapter is focused on the development of a monitoring algorithm devoted to detecting and isolating an open circuit of one of the electric motor’s phases. Starting from the definition of its required performance, a logic flow in Stateflow environment is developed to implement the intended function using a fault detection section and a fault isolation section. The first one is responsible of continuously monitoring the health of the system through the measurement of currents converging in the neutral point. The detection is triggered when a displacement beyond a certain threshold from the expected zero signal is identified, persisting over a time window. The fault isolation section is thus activated revealing the faulty phase and isolating the entire engine segment, switching the system to a fault-operative mode. Besides, an investigation is carried out to show how different types of faults, namely open circuit of a phase, short circuit of a phase, hardover of required current of an engine module and faulty angular speed sensor. These fault modes affect the overall performance and the behaviour related to various signals generated by different components. This work is intended to establish a base to develop a more complete algorithm capable of detecting other types of faults in different operative scenarios, exploiting at its best redundant features present in the TERSA architecture