Neural network controller for microturbine power plants

Microturbines (MTs) as alternative energy sources demonstrate high potential to meet user's needs for distributed generation application. Efficient model and controller are essential to understand and investigate the MT operation characteristics. This paper introduces a controller based on artificial neural network (ANN) for stand alone MT power plants. In addition, the paper presents a comparison between the performance of the MT when using traditional PI and ANN controllers. MT has three control loops, namely; the power, the temperature, and the speed control loops. In addition, in this paper the MT is connected to a synchronous generator (SG) which includes a voltage control loop. An ANN with four inputs and four outputs is used to replace the four PI control loops in the MT and the SG models. Three test measures have been used to compare the performance of the MT with PI and ANN controllers, absolute mean error (AME), root mean square error (RMSE), and standard deviation error (SDE). Test results of MT with the ANN controller indicate high performance compared to the conventional PI controller.