Publication: Investigation on modifying grasshopper optimization algorithm (goa) for function optimization
Date
2022-07-01
Authors
Ismail, Muhammad Aqil Uzair
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Optimization is a very important process involved in most of the systems in this world. An example, the volume of fuel burnt in a car engine is adjusted by the ECU depending on the car’s speed. Optimization algorithms can be categorized into two types that are deterministic methods and stochastic methods. Due to the limitations of the classical optimization algorithms, many metaheuristics based on stochastic have been introduced to solve optimization problems. Among them is swarm intelligence algorithm which is actively being explored to this day and continuously being developed to get the best optimization algorithm. In 2017, grasshopper optimization algorithm (GOA) has been introduced. It is a simple yet powerful algorithm that have been applied in various process such as data clustering, feature selection and image processing. Despite that, GOA still have limitations when the optimization problems is complex which are poor local minimum avoidance and premature convergence. This are because of the linear comfort zone coefficient c. Because of that, many variant of GOA have been introduced to minimize the limitations existed and thus expand its application. Therefore, this research proposed three GOA variant with non-linear
comfort zone coefficient c. The proposed algorithms have been tested on 23 benchmark functions and 25 functions in CEC2005. The results of the GOA variants have been compared to the standard GOA. Besides, an overall comparison between each GOA variant and the standard GOA has been made. The overall comparison shows a little improvement in term of global minimum achievement and convergence speed for 25 CEC2005 benchmark functions.