Abstract

Automatic generation control (AGC) is required by power utilities to match precisely the power demand at any given time. The AGC system is designed to achieve the short-term reallocation of power to keep the change in system frequency within set pre-specified limits. Tie-lines power flow control is another important objective of AGC. System frequency is a good indication of an imbalance between power generation and load. An imbalance between generated and demanded power will be reflected in a change in frequency. This will result in an increase or decrease in the speed of turbine- generator set. In turn, system frequency will deviate from the nominal frequency. Different controllers can be employed to maintain frequency and tie-line power at set values. An inappropriate controller gain value can affect the system performance and even it can become an unstable system. Therefore, it becomes essential to tune the controller parameters to achieve desired control performance. Many researchers have proposed a number of control methodologies like auto-tuning, self-tuning, etc. These conventional methods have some disadvantages, which may be easily overcome by employing evolutionary algorithms. In the past few years, many researchers modified particle swarm optimization (PSO) algorithms and all such changes have led to gradual convergence, but it still requires a large number of runs to achieve a minimum value for a multi-dimensional problem. The proposed work introduces a new variation in classical PSO, in terms of statistical parameters, called hybrid statistically tracked particle swarm optimization (hybrid STPSO), to search for the global best value at an accelerated convergence level. Basic PSO can enmesh at the local best value in the multi-dimensional shifted, rotated problem, yet no subsequent changes in particle velocity are observed. A mathematical parameter at this stage offers a new direction to the best global value. The proposed algorithm uses properties of statistical parameters to accelerate the velocity of particles to find the best possible value with fast and improved convergence speed. The effectiveness of the proposed algorithm for frequency control of the system with and without the effect of dead band is analyzed by comparing its performance with other evolutionary algorithms like PSO, chaotic PSO, differential evaluation, big bang–big crunch, and teaching–learning-based optimization. A significant improvement in the performance of hybrid STPSO as compared with other variants of PSO and different evolutionary algorithms is observed in the frequency control of the AGC system.

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call