Numerical and Neural Network Modeling and control of an Aircraft Propeller

In this paper, parametric and numerical model of the DC motor, connected to aircraft propellers are extracted. This model is required for controlling trust and velocity of the propellers, and consequently, an aircraft. As a result, both of torque and speed of the propeller can be controlled simultaneously which increases the kinematic and kinetic performance of the aircraft. Parametric model of the motor is derived by conducting standard tests such as locked rotor test and step and sine wave input one. In order to derive a neural network and numerical model, a set of sinusoidal, triangular, and random step signals are applied as the input to the motor and its speed is recorded as an output. Neural network of the motor is extracted by using these datasets and considering a multilayer perceptron (MLP) neural network structure with Levenberg-Marquardt training method. Results of the numerical model and parametric model are compared and validated by experimental implementations. The superiority of the proposed method is also shown respect to traditional PID algorithm.