Global Optimization with Petal Guided Flower Pollination Algorithm

Heuristics are known to infuse various search procedures with intelligence. Metaheuristic Algorithms are general purpose heuristics that are in general inspired by nature. There are several phenomena in nature that can be studied and utilized to solve different types of decision-making problems in engineering and other domains. It is definitely an interesting domain of research where the modeling of natural processes is done in mathematical context to extend further support of computing prowess to solve various types of problems. The novel algorithm named as Petal Guided Flower Pollination Algorithm, proposed in this paper, is modeled on the Flower Pollination approach and is specifically intended for global optimization purpose. Benchmarking functions are typically used to analyze optimization algorithms. The focus is on function optimization where benchmarking functions have been used. The comparison of this new algorithm has also been carried out with the standard Flower Pollination Algorithm. The Petal Guided Flower pollination algorithm does work well in the context of several benchmarking functions. A very basic statistical analysis has also been carried out for both algorithms to gain better insight. Evolution curves along with measures of central tendency have been used in this context.