Discussion On The Method Of Determining Parameters Contained In Particle Swarm Optimization Algorithm

In practical and scientific computing,many models will contain many parameters.The selection of these parameters will greatly affect the calculation of the model,and sometimes the multiple models will be used iteratively so that the final calculation will be more.In the intelligent optimization algorithm,the selection of parameter values will also directly affect the calculation of the algorithm.Because of the randomness of the algorithm itself,the calculation of the algorithm is also uncertain,then the problem of finding the minimum value of the calculation is a random optimization problem.Therefore,in order to reduce the amount of computation,the selection of parameter values is particularly important,and also has practical value and practical value.In this paper,we take particle swarm optimization algorithm as an example to discuss the selection of parameter values.The stochastic optimization problem is also a process of random search.In this paper,we discuss the optimization of parameters in particle swarm optimization by two aspects of the characteristics of random variables and the expected value model in random programming.The main research work is as follows:1.Because of the large range of parameters,first of all,we should consider narrowing the range of search,searching the most promising areas by searching the tabu table and the hope table,then searching the most promising areas,so as to determine the optimal scope of research.2.In view of the characteristics of random variables,it is necessary to solve the target function corresponding to the parameters,but the analytic formula of the objective function is not easy to write.Then,the variance analysis method,which has the statistical inference of the random variables in mathematical statistics,is used to judge the degree of the influence of the parameters,and then the two approximation is used to find the eyes.In order to solve this problem,a method of variance analysis with two approximation is put forward to solve this problem.In this way,the thought and basic steps of the variance analysis method for the one,two and three yuan two approximation are given.3.The expected value model is given in random programming,but because the analytic expression of the objective function is not easy to write out,the approximate value of the objective function can be used instead of the objective function value.The traditional algorithm and intelligent algorithm are usually used in the model solution,but for this kind of special stochastic optimization problem,this paper uses the improved traditional algorithm with two approximation of the Powell algorithm and particle swarm optimization to discuss the problem of parameter value.4.Using the above two approximation method of variance analysis,the Powell algorithm with two approximation and the particle swarm optimization algorithm,the parameters of the particle swarm optimization are optimized.The example shows that the method proposed in this paper is feasible.