Efficient line-search modified bat algorithm for solving large-scale global optimisation problems

An efficient line-search modified bat algorithm (EMBA) is proposed to solve large-scale global optimisation problems. A balance between exploration and exploitation abilities is achieved. Firstly, a line search to an accurate step size of a particle towards the global optimum is presented. The generated step size depends on the proximity of the particle to the global optimum and it is directly proportional to the dimension of a problem. This proportion makes EMBA capable to handle the high probability of an explosion in the initial values of objective functions in large-scale optimisation problems. Secondly, the velocity of a particle is clamped within pre-defined boundaries and penalised, if necessary, to ensure that both the velocity and position of a particle are within their boundaries. These modifications combined make EMBA able to converge to the global optimum in a few iterations. The experimental results show the efficiency of EMBA when comparing with well-established algorithms.