Abstract: In engineering optimization,most problems are continuous optimization problems,that is function optimization problems.Aiming at the problems of slow convergence speed,low precision and easy to fall into local optimization in the later stage of Cuckoo algorithm,this paper proposed an improved Cuckoo search algorithm based on the logarithmic decline of nonlinear inertial weights and the random adjustment discovery probability.Firstly,in the update formula of the path and position of the cuckoo homing nest,a update method that inertia weight decreases nonlinearly with the number of evolutionary iterations is designed to improve the nest location and coordinate the abilities of exploration and exploitation.Secondly,the discovery probability with random adjustment is introduced to replace the discovery probability of fixed value to make the larger and smaller discovery probabi-lity appear randomly,which is beneficial to balancing the global exploration and local exploitation of the algorithm,accelerating the convergence speed,and increasing the diversity of the population.Finally,the logarithmic decreasing parameter and stochastic adjustment discovery probability are analyzed and tested,and the optimal parameter combination of logarithmic decrement and the optimal range of stochastic adjustment discovery probability are selected.At this time,the optimization effect of the function is the best.Compared with other evolutionary algorithms (BA,CS,PSO,ICS),DWCS greatly improves the precision of optimization,significantly reduces the number of iterations,and effectively improves the convergence speed and robustness.In 16 test functions,DWCS can converge to the global optimal solution,which proves that DWCS has a strong competitive power in solving the optimization problem of continuous complex functions.

Key words: Cuckoo search algorithm, Discovery probability, Function optimization, Logarithmic decreasing, Parameter selection