Implementation and comparison of modified binary particle swarm optimization approaches to an interruptible load dispatch problem

Abstract

The nature of interruptible load programs- immediate, variable, and complex- set it apart from most dispatch type optimization problems. Not only does one have to minimize the total cost of interruption, but also has to generate a dispatch that follows time-based operational limits, and sometimes minimize the frequency of interruptions made. Previous studies have shown the promise of binary particle swarm optimization in obtaining a dispatch that operates within constraints. However, although much improvement has been demonstrated and tested on the binary particle swarm optimization method itself, some are not yet tested in solving an interruptible load dispatch problem. The study explored recent published improvements to the binary particle swarm optimization algorithm to see if making changes to the way particles update and communicate will help the swarm find more optimal solutions. The study found that using the canonical binary particle swarm still provides the most optimal results of those evaluated. The study also determined that the essential version of BPSO also holds great potential in obtaining optimum solutions while ensuring that particles update according to their trajectory.