Adaptive/optimal neurocontrol based on adaptive critic designs for synchronous generators and FACTS devices in power systems using artificial neural networks

The objective of this research is to design nonlinear optimal controllers based on the Adaptive Critic Designs (ACDs) algorithm using artificial neural networks (ANNs) for the control of synchronous generators and flexible alternating current transmission system (FACTS) devices in an electric power grid.; In the first stage of this study, the on-line identification and indirect adaptive control methodology are described for the design of a new controller as an alternative to the conventional linear controllers (CONVC) to control a synchronous generator, using the two different types of ANNs, namely a multilayer perceptron neural network (MLPN) and a radial basis function neural network (RBFN). Then, the research is expanded to design an optimal neurocontroller based on the heuristic dynamic programming (HDP) and dual heuristic programming (DHP) algorithms among the family of ACDs for the control of a synchronous generator. In both the indirect adaptive neurocontrol and HDP/DHP based optimal neurocontrol schemes with the MLPN and RBFN, the dynamic control performances of the proposed controllers are compared.; Among the series FACTS devices as the emerging technology, the static synchronous series compensator (SSSC) and the series capacitive reactance compensator (SCRC) are investigated. A novel internal optimal neurocontroller for the SSSC and a new external optimal neurocontroller for the SCRC are developed based on the DHP optimization algorithm. Finally, the DHP optimization algorithm is applied in the design of two optimal neurocontrollers; one is for the control of a synchronous generator and the other is for the external control of the SCRC FACTS device on a practical large-scale multi-machine power system network. The system performances, as affected by interactions of the two DHP optimal neurocontrollers with different control objectives, are evaluated.