Automatic Generation Control of Multi-Area Interconnected Power Systems Using Hybrid Evolutionary Algorithm

Automatic Generation Control of Multi-Area Interconnected Power Systems Using Hybrid Evolutionary Algorithm

Omveer Singh (Maharishi Markandeshwar University, India)
DOI: 10.4018/978-1-5225-2128-0.ch010

Abstract

A new technique of evaluating optimal gain settings for full state feedback controllers for automatic generation control (AGC) problem based on a hybrid evolutionary algorithms (EA) i.e. genetic algorithm (GA)-simulated annealing (SA) is proposed in this chapter. The hybrid EA algorithm can take dynamic curve performance as hard constraints which are precisely followed in the solutions. This is in contrast to the modern and single hybrid evolutionary technique where these constraints are treated as soft/hard constraints. This technique has been investigated on a number of case studies and gives satisfactory solutions. This technique is also compared with linear quadratic regulator (LQR) and GA based proportional integral (PI) controllers. This proves to be a good alternative for optimal controller's design. This technique can be easily enhanced to include more specifications viz. settling time, rise time, stability constraints, etc.
Chapter Preview
Top

Evolutionary Algorithms

Evolutionary algorithm is a gradually enhancing research discipline taking soft computing techniques that are motivated by idea of natural evolution. The three prime mechanisms that move evolution forward are reproduction, mutation, and natural identification (i.e., survival of the fittest theory which is published in the Darwinian principle). In the biological practice, these mechanisms reflect life frameworks to a special environment over successive productions. EAs take these mechanisms of natural evolution in simple paths and breed progressively better results to a wide variety of complex optimization and design issues. An EA uses some mechanisms motivated by biological evolution: regeneration, mutation, recombination and selection. Further, EA categorizes in GA, genetic programming and evolutionary programming etc. However, GA technique is more probabilistic methodology than the other EA strategies.

J. H. Holland was the first to introduce GA (Holland, 1992) and its Pseudo-code is represented in the below.

  • GA procedure

  • begin

  • completion:=worse

  • Initialisation {initial population}

  • while (No completion) do

  • Evaluation

  • if (No termination condition) (fitness functionperformedon chromosomes)

  • then

  • Selection {intermediatepopulation}

  • Crossover & mutation operation (nextpopulation)

  • else

  • completion:=good

  • end

  • end

Complete Chapter List

Search this Book:
Reset