Soft computing techniques for HVDC control

Abstract

High Voltage Direct Current transmission systems have a very important place in today’s power system Scenario. HVDC link is inherently non-linear in nature. Obtaining an accurate mathematical model of the system is not possible. Power transmitted in a HVDC plant depends upon the efficiency of the controller and is the most important and complex part. Traditionally Proportional plus Integral Controllers are employed for rectifier pole current control and inverter side extinction angle control but they suffer from the problems of fixed gains and can operate well over a limited operating range. So, this thesis deals with enhancing the performance of PI controllers for HVDC by employing the intelligence of expert techniques available in literature which can deal with non-linear optimization problems. In the first part of the thesis, a hybrid technique incorporating the features of Fuzzy Logic and Neural Networks to self-tune the PI controller is implemented. The error in the value of current and rate at which this error varies were considered as the base parameters. In the second part of the thesis a Genetic Algorithm-Neural Network technique is adapted to self –tune PI controller gains. Results from Simulations illustrate the potential of the proposed control scheme as GA-NN technique successfully adapts to different system conditions and is able to minimize the total current error. Frequently occurring transient stability because of severe power system disturbances can be overcome by interconnection of HVDC links with interconnected systems.