New design of an intelligent electromagnetic torque controller based on neural network and fractional calculus: variable-speed wind energy systems application

Abstract

To achieve high efficiency of the electricity production for wind turbines, an improved controller has a crucial role in achieving this goal by capturing the most wind energy based on the maximum power point tracking approach. This research paper suggests a new nonlinear controller design to control the electromagnetic torque for horizontal-axis variable-speed wind power with three blades connected to the grid to render them more profitable and efficient in terms of the highest rate of electricity production. This proposed controller binds the sliding mode (SM) with the fractional calculus (FC) and the neural network (NN) to exploit the benefits of each technique. The SM is a popular technique and is the most used in controlling nonlinear systems. The effectiveness of the SM is shown in its ability to stabilize the system, drive it to the desired state in a finite time, and reduce the sensitivity to parameter variations. However, the main drawback of SM is the chattering phenomenon. This phenomenon refers to the high-frequency oscillations that occur around the sliding surface. The chatter arises due to the discontinuous nature of the SM control law, which switches between different control actions to keep the system states on the sliding surface. The main contribution of this present work is to tackle this undesirable issue that can damage the system, destroy the components, and lead the system to instability. The solution lies in suggesting a new controller that combines SM, FC, and NN because the FC provides better modeling concerning the dynamic behavior by outperforming the classical operators by using the non-integer order. And, the NN aims to estimate the unknown dynamics that are incorporated in the equivalent term in SM, reduce the chattering by compensating for the uncertainties, allow the system to adjust to varying conditions related to the uncontrollable wind, and make an adaptive controller by improving its performance over time by learning the system dynamics. This proposed integration between SM, FC, and NN gives a good performance that showcases via emulation results under three different scenarios of wind speed. In addition, in each scenario, two tests are performed to prove the effectiveness of the suggested law control.