Improved Krill Herd Algorithm And Its Application In Structural Reliability Analysis

Krill herd algorithm(KH)is a new meta-inspired optimization algorithm.Gandomi first proposed KH in 2012,inspired by Antarctic krill herds foraging for food.The krill herd algorithm has become a research hotspot due to its strong exploitation,fewer related parameters to adjust,and easy implementation.Although KH has some better performance than other meta-heuristic algorithms in some optimization problems,there are still some disadvantages,such as lack of an optimal strategy for selecting the initial krill swarm population distribution,the balance between exploration and exploitation capabilities,global convergence slow and other disadvantages.Based on the krill herd algorithm theory and implementation process,this paper proposes two improved versions of krill herd algorithm,namely Target-Oriented Krill herd algorithm and Cooperation and Competition Krill Herd Algorithm,and their application in solving the structural failure probability.In the research of krill herd algorithm:To solve the slow global convergence of krill herd algorithm,a cooperative and competition krill herd(CCKH)algorithm is proposed.The cooperation is making the optimal krill and food cross to generate a better candidate solution before the krill swarm motion calculation.Competition is a strategy that selects the better value between optimal krill and virtual food to guide the movement of krill swarm.Finally,the optimization efficiency of Levy flight strategy is better than physical diffusion.Therefore,use Levy flight strategy instead of physical diffusion to accelerate the global convergence of the algorithm.Aiming at balance the exploration and exploitation of krill herd algorithm and changing the initial distribution of krill swarm,a Target-oriented krill herd(TOKH)algorithm is proposed.TOKH modifies the initial distribution of krill swarm,increases the distribution density of optimal krill and virtual food.Then cross optimal krill and virtual food location to generate a new candidate solution to further improve the efficiency of optimization.In terms of engineering application research:Combining the TOKH algorithm and CCKH algorithm with the structural reliability first order reliability method(FORM),and by adding a penalty function,the constrained optimization problem is transformed into a general optimal problem.Thereby the Hasofer-Lind reliability index is solved.The numerical results show that the application of TOKH algorithm and the CCKH algorithm can not only solve the structural failure probability with high accuracy,but also directly solve the structural failure point.Compared with Monte Carlo simulation,the time required to solve the problem is also greatly shortened,and the efficiency of solution is greatly improved.So,in this paper,the improvement measures not only improve the optimization performance,accelerate the global convergence,but also improve the efficiency of solving the structural reliability index,and provide new methods to solve structural reliability problem.In engineering,it also has better practicability.