Application of Neural Networks to Load Frequency Control in Power Systems with Four Control Areas

Rapid population growth and technological development is a 21st century phenomenon. This phenomenon eventually increases the demand for electricity and its supply reliability. Power systems are complex networks consisting of generation, transmission and distribution of electricity to customers over a large geographical area. Power systems are interconnected to enable a secure and economical supply. Automatic Generation Control (AGC) or Load Frequency Control (LFC) is a very important subject in power systems for a reliable and quality of electricity supply to costumer. Load Frequency Control helps to reduce deviations during transient processes by moving the error to zero value in steady state. The main objective of ACG in interconnected systems is to maintain the frequency at nominal values of the desired generation power output. There are various control techniques that have been applied to power and frequency control problems. The PI controller is among the simplest to implement but the set-up time is relatively large and causes a lot of oscillations in the frequency response. The best alternative to the PI controller is the widely used Fuzzy logic controller. However, this controller has its limitations as it has good dynamics only when defining the correct number of pointer functions. To enable an even better power and frequency tuning the controller with artificial neural networks (ANN) has found great use because it has faster control than other types and can improve transient response through training/learning procedure. This paper addresses the performance evaluation of the ANN controller in controlling power and frequency signals in four area interconnected system with a combination of thermal � thermal, hydro � thermal and hydro � hydro generation. It analyzes the performance of frequency response. The performance is estimated by comparing the result of PID controller with ANN controller