This research project deals with the application of Artificial Neural Network (ANN) based Adaptive Neuro-Fuzzy Inference System (ANFIS) approach to Decentralized Automatic Generation Control (DAGC) Scheme for interconnected multi-area Power System. The proposed ANFIS Controller combines the advantages of Fuzzy Controller as well as quick response and adaptability nature of ANN. The design controller for Decentralized Automatic Generation Control has three major objectives i.e., To maintain the system Frequency at nominal value (50Hz), To maintain the net Tie-line power interchange from different areas at their scheduled values and To incorporated various types of possible transactions such as PoolCo-based transactions, Bilateral transactions or a combination of these two.
Conventionally, for issues related to Automatic Generation Control (AGC), the frequency deviation is minimized by the flywheel type of governor of synchronous machine.However, the significant control is not achieved for the Load Frequency Control (LFC) objective. In this context, the supplementary control is introduced to the governor via signal directly proportional to the frequency deviation plus its integral action. The proposed approach with non-interaction between frequency and tie-line power control and each control area responsible for its own load variations. The technique based on coordinated system-wide correction of time error and inadvertent interchange is incorporated for AGC (generally referred to as Area Control Error (ACE)).
In a Practical power system, there may be more than two areas, and each of the areas may have different ratings. Contrary to the centralized control for a large scale power system, Decentralized control is preferable, because it reduces the computational burden with pass of the communication between different systems and make the control more feasible and simple. n order to overcome the problem arising out of the centralized control, the decentralized control approach has been addressed. The basic objective of later technique is to make the composite ystem divided into subsystem, each of which control separately. The advantage of a decentralized controller is to reduce complexity and therefore, make its implementation more practical.
The prominent feature of fuzzy and neural network based schemes is that they provide a model-free description of control systems and do not require model identification. In this research a control scheme based on ANFIS, which is trained by the results of off-line studies, obtained using genetic algorithm has been proposed. ANFIS is considered to be an adaptive network, which is very similar to neural networks. Adaptive network has synaptic weights, but has so called adaptive and non-adaptive nodes. It must be said that adaptive network can be easily transformed to neural networks architecture with classical feed no forward topology. This adaptive network is functionally equivalent to a fuzzy inference system (FIS). Using a given input/output data set, ANFIS adjusts all the parameters using back propagation gradient descent and least squares type of method for non-linear and linear parameters respectively.
The proposed Decentralized Automatic Generation Control (DAGC) based on ANFIS Controller scheme has been simulated on a practical 39-bus New England system and a 75-bus Indian power system. The performance of the Decentralized Automatic Generation Control (DAGC) ANFIS controller is compared with the results of Decentralized Automatic Generation Control (DAGC) based integral controller. Simulation results indicates that the controllers exhibit better performance.