Nature-Inspired (NI) Toolbox is a Particle Swarm Optimization (PSO) based toolbox which is developed in the MATLAB environment. It has been released under General Public License and hosted at SourceForge.net (http://sourceforge.net/projects/nitool/). The purpose of this toolbox is to facilitate the users/designers in design and optimization of their systems. This chapter discusses the fundamental concepts of PSO algorithms in the initial sections, followed by discussions and illustrations of benchmark optimization functions. Various modules of the Graphical User Interface (GUI) of NI Toolbox are explained with necessary figures and snapshots. In the ending sections, simulations results present comparative performance of various PSO models with concluding remarks.
TopDownload NIT.zip folder from SourceForge.net (http://sourceforge.net/projects/nitool/) and follow these steps to embed it in the MATLAB environment:
- 1.
Unzip the downloaded NIT.zip file. This contains two folders (a) nitool folder – copy into the /matlab/toolbox and (b) NI Toolbox folder – contains demo and make it your Work Directory.
- 2.
In the MATLAB environment click Start → Desktop Tools → Path → Add Folder
- 3.
Specify the nitool folder path as /matlab/toolbox/nitool, Save and Close.
- 4.
Now, NI Toolbox is ready to use. Type nitool in the MATLAB Command Window to run NI Toolbox for a fresh system description.
TopParticle Swarm Optimization
PSO belongs to the category of Swarm Intelligence (Kennedy and Eberhart, 2001) tool and is useful in solving global optimization problems. It was originally proposed by James Kennedy, as a simulation of social behavior, and was introduced as an optimization method in 1995 (Eberhart and Kennedy, 1995, Kennedy and Eberhart, 1995). PSO is an evolutionary computing technique related to artificial life, specifically to swarming bodies, as it involves simulation of social behaviors.
PSO implementation is easy and computationally inexpensive, since its memory and CPU speed requirements are low (Eberhart et al., 1996). Moreover, it does not require gradient information of the fitness function but only its values. PSO has been proved to be an efficient method for many global optimization problems and, in some cases, it does not suffer from the difficulties experienced by other evolutionary algorithms (Eberhart and Kennedy, 1995).
What differentiates the PSO paradigm from other instances of evolutionary computing is memory and social interaction among the individuals. In the other paradigms, the important information an individual possesses, usually called genotype, is its current position, however, in PSO, really important asset is the previous best experience. Each individual stores the best position, found so far, that drives the evolution toward better solutions.