Application of Artificial Neural Networks for Monitoring and Optimum Operation Prediction of Solar Hybrid MGT Systems

Abstract Hybrid energy system consisting of a parabolic dish solar concentrator and a micro gas turbine (MGT) has been considered as promising distributed generation technology, since it can be operated as a stand-alone system for power or combined heat and power (CHP) applications in remote areas with no connection to the grid. The main concern when it comes to distributed generation is the ability of maintaining high availability. Therefore, given the intermittency of the solar resource, the availability of consistent and computationally fast tools for modelling and monitoring of solar micro gas turbines is essential for adequate and optimum operation. This paper presents the application of artificial neural networks (ANNs) for performance prediction and monitoring of a hybrid solar MGT system. For this purpose, a validated in-house tool, developed for evaluating the performance of solar-hybrid MGT, was used to generate simulated data at various operational conditions by varying solar irradiation and ambient conditions. The obtained data was used to train the ANN model. The prediction accuracy of the ANN model was tested using a data set, which were not used during the training process. The results showed that the ANN model can predict the solar hybrid MGT performance with high accuracy and could serve as an accurate baseline model for monitoring applications. Finally, the developed ANN model was integrated with an optimization algorithm. A case study was conducted using the developed ANN model for multi-objective optimization of the hybrid solar MGT. By varying turbine inlet temperature and rotational speed, the system performance at part load operation were analysed resulting in a Pareto front for maximum electrical efficiency and minimal operational cost. Multi-objective genetic algorithm (GA) based on controlled elitism concept was applied to find the optimum solution.