A Pigeon-inspired Algorithm For Function Optimization Problems

With the development of science and the emergence of new technology,a lot of difficult problems in our daily life are generally solved.However,after the method has grown out of nothing,solving problems efficiently has become a new target continuously pursued and investigated by domestic and foreign scholars.In the realistic world,Optimization problems are characterized as large scale,nonlinear and non-convexity which are always NP-hard.Therefore,the function optimizations,especially the computationally expensive optimization problems which are treated as the abstract of real problems are becoming a hotspot issue of researches.Currently,many swarm intelligence optimization algorithms have been applied to solve the function optimization problems,which illustrate the effectiveness of swarm intelligence optimization algorithms aiming at the continuous optimization problems.In this paper,we intend to solve the function optimization problems based on the pigeon-inspired algorithm proposed by Duan(2014),which was initially applied in the aerospace field.However,this algorithm has disadvantages of poor precision,low convergence speed and narrow application field due to its short development history.For these disadvantages,we developed a modified pigeon-inspired algorithm to improve the strategy in different stages of the algorithm to increase algorithm efficiency.Firstly,chaotic and opposite initialization were used in the initializing process to ensure the diversity of solutions and speed up the convergence.Next a redistributing strategy with Cauchy mutation was introduced in the iterative process to jump out of local optimum in some extent.Finally,self-adaptive parameters were adopted to balance the global search and local search,which can increase the convergence speed.Experimental results of twenty test cases indicated that the modified pigeon-inspired algorithm is more effective than the original pigeon-inspired algorithm,CLPSO and CMAES algorithm.