The Improvement And Application Of Krill Herd Algorithm

Krill Herd Algorithm(KH)is proposed in 2012 by Gandomi and Alavi.KH has strong abilities of convergence and robustness,and it is easy to be coded.But KH has some disadvantages,such as poor convergence accuracy,low computational efficiency,less applications and so on.Further studies,improvements and applications are be done in this paper,the main works are as follows:1.An improved krill herd algorithm based on natural selection and random disturbance is proposed.This improved algorithm firstly applies nonlinear decreasing strategy based on time of induced weight and foraging weight into KH.And then random disturbance is added into the process of generating the new generation of krill herd population.And the natural selection mechanism enhances the quality of the individuals in the krill herd population.Those steps can effectively balance the abilities of global search and local exploration.2.A hybrid algorithm based on improved particle swarm optimization and krill herd algorithm is proposed.A new nonlinear decreasing strategy of foraging weight and induced weight is added into KH.And then the particle swarm optimization with exponential decreasing inertia weight is involved into the improved KH.The strategy of double sub-populations is applied to the iterative process to share the useful information and improve operating efficiency.The mechanism of survival of the fittest is introduced lastly.The results of experiments demonstrate that the proposed algorithm has a significant advantage in the convergence accuracy and operating efficiency.3.Puts the constraint handling mechanism into the hybrid algorithm to solve the constrained optimization problems,Nonlinear Mixed Integer Programming Problem.A comparison of the different algorithms about the convergence accuracy and the success rate is shown.