Research On Group Intelligence Algorithm Search Analysis And Portfolio Problem

Group intelligence is a general term for a class of algorithms.It refers to human beings inspired by the behavior of various biological groups in nature.By mimicking these intelligent behaviors,they find better strategies for optimizing.With the advancement of science and technology,some traditional optimization methods have been unable to effectively solve complex and ever-changing problems in reality,especially the NP-hard problem.For this reason,more and more scholars have begun to adopt multi-group swarm intelligence optimization algorithms to solve the problem.As a result,they show a significant advantage,especially in the field of engineering applications and scientific research,swarm intelligence algorithms are getting more and more attention from scholars.Espezua and other scholars proposed eight(four groups)cross-genetic operations,and gave qualitative results for the size of the search area that each pair of operations may cover,but there is no quantitative proof.In this paper,the quantitative analysis of the two pairs of real-coded cross-operations is carried out,and the analysis results of the comparison of the size of the two search areas are given.It is proved that the new cross?operation and the corresponding original cross-operation are conserved.At the same time,it has a relatively wide neighborhood,which theoretically proves that genetic algorithms have the inherent reasons for maintaining group diversity and better algorithm performance.Aiming at the shortcomings of differential evolution algorithm,such as slow convergence rate and easy to fall into local optimal solution,this paper proposes a differential evolution algorithm integrated into cluster analysis.Firstly,the clustering analysis method is used to cluster the populations of the differential algorithm,extract the representative individuals,replace the poor individuals in the original population with new individuals,and remove the redundant information in the population to optimize the population.This allows the entire population to converge quickly and accurately to the global optimal solution.Finally,the simulation experiment is carried out by using MATLAB.The simulation experiment is carried out on the CEC2005 test function library.The results show that the differential evolution algorithm with cluster analysis strategy can not only effectively suppress premature convergence,improve the convergence speed,but also be simple and efficient.Strong and other characteristics.The portfolio problem is a nonlinear programming problem,and the traditional algorithm cannot effectively find the optimal solution.In this paper,the new cluster-based difference algorithm is used to solve the mean-VaR model.The external penalty function method is used to deal with the inequality constraints in the model,which is transformed into an unconstrained optimization problem that is easier to solve.50 stocks in Yahoo Finance are selected.The empirical analysis shows that the algorithm has achieved good resullts in solving the portfolio problem.The results of the solution meet the investment objectives and constraints,and reflect the different types of income risk requirements among investors.And has good practicality.