Article Preview
TopIntroduction
In real life, optimization is mainly considered for highly nonlinear complicated problems, with complex constraints and different parameters of design. The basic aim of optimization problem is commonly affiliated with the decrement of cost, time, or increment in profits, benefits and performance (Mazhoud, Hadj-Hamou, Bigeon, & Joyeux, 2013). The standard deterministic algorithms practically, are unable to tackle a huge number of problems, particularly, when it comes towards the multimodal objective functions with distinct local optima. At that time, scientists tried to move towards the nature for finding new ideas that will have enough ability for problem solving. Many algorithms are introduced in the search of new idea, which are used for function optimization such as genetic algorithm, evolutionary programming, differential evolution and many more (Jr., Fister, & Yang, A Hybrid Bat Algorithm, 2013).
In the most recent decades, popular researchers have seen the advancement in nature stimulated algorithms that was consider as the most arising algorithms to determine the optimization problems. The subarea of Computational knowledge is known as Swarm Intelligence (SI), which is devoted to echo. The behavior of echo is incorporated to figure out the solutions that are not able for producing by the optimization approaches (Kabir, Sakib, Mustafizur, & Shafiul, 2014). The algorithms that are based on the SI optimization started with initializing the population into the multi-dimensional search space through several methods (Bansal, 2017). With these characteristics of SI many algorithms are introduced and applied in various fields like particle swarm optimization, ant colony, monkey algorithm (Zhao & Tang, 2008), glowworm swarm optimization algorithm (Krishnanand & Ghose, 2009) and so on. A SI algorithm is capable to solve a large number of complex optimization problems and also gives superior results, where many other algorithms face difficulties. Therefore, the area of its applications is growing rapidly (Li & Zhou, 2014).
Moreover, another novel swarm methodology called Bat algorithm (BA), proposed by X. Yang in 2010 (Wang, Chang, & Zhang, A Multi-Swarm Bat Algorithm for Global optimization, 2015). In this methodology, the echolocation bats was simulated with deviating loudness and pulse rate (Yang & Gandomi, 2012). The main concept of bat’s echolocation is to provide a hunting technique (Alihodzic & Tuba, Improved Hybridized Bat Algorithm for Global Numerical Optimization, 2014). Gradually, researchers attracted towards the BA and presented variety of applications, like path planning problem, loud frequency control, IIR system identification and Unit Commitment problem (Cui, Li, & Kang, 2015).
Research described that micro bats work with time, from detection and emission of echo a time delay is used. The time divergence between their ears and the distinctions of echo’s loudness mapped the 3-D scenario of the environment. They can identify the distance, direction, the quality, and even the moment speed of target. However, a small number of bats have sharp eyesight, but mostly have very powerful smell sense. In practical, Bats use the combination of their senses to identify their target efficiently with smooth navigation. Although, here we have only concern with the echolocation behavior of micro bats (Yang & Gandomi, 2012).
There are two most critical components in the advanced meta heuristic algorithms are: exploration and exploitation. An investigation ability that can find many search spaces to bring out the global optimum solution is called exploration. While on the other hand, a method that reveals the optimum solution with the help of available best solution is known as exploitation. An algorithm can outperform, when these components are well-balanced. A lower amount of exploration, while a large amount of exploitation can cause the algorithm to stuck into the local optimal solution. Meanwhile, a large amount of exploration and too little exploitation could affect the convergence speed and also the performance of an algorithm. According to the some researches, it is a good practice to adopt exploration at first and then exploitation for better performance of algorithm (Tan, Chiam, Mamun, & Goh, 2009) (Yılmaz, Kucuksille, & Cengiz, 2014).
The objective of this study is to enhance the swarm diversity of the BA’s for improving the exploration and exploitation capability of the standard BA. Exploration is process of searching prey into the wide area of search space, while exploitation refers the search in the local area. To maintain a balance between these two properties there must be robust population initialization approach.