Research And Application Of Two Algorithms Of Particle Swarm Optimization And GM(1,1) Model

Particle swarm optimization is a new artificial intelligence optimization algorithm derived from simulating the ability of group biology to work together to optimize each other.The clustering problem occupies a large proportion of data mining and has a great deal of research and application in many fields.The clustering problem also belongs to the optimization problem,the combination of particle swarm optimization and clustering method has practical significance.There is a problem that when particle swarm optimization is combined with K-means algorithm,that the particles flying out of the cluster space are constrained by the displacement and the particles are fixed on the boundary,which leads to the error clustering.Limiting Boundary of Particle Swarm Optimization is a algorithm to solve this problem.The equations of particle swarm system are simplified,and the position difference equations of particle swarm are solved.The conditions for the parameters in particle swarm optimization to be satisfied are obtained.The LBPSO algorithm is used to carry out the clustering experiments.The LBPSO algorithm is applied to Iris dataset,Wine dataset and Glass dataset respectively.By comparing the experimental results,the feasibility of the algorithm is verified,which shows that the algorithm can improve the accuracy of clustering.Grey system is the main subject of this discipline,which is used as prediction,control and decision-making.In order to improve the simulation effect and prediction accuracy of non-equidistant GM(1,1)model in grey system,the idea of innovation is combined in the traditional cumulative non-equidistant GM(1,1)model.The initial conditions of the cumulative non-equidistance GM(1,1)model's and its background value are also improved.The valid range of the parameters are discussed.And the theorem is obtained and proved.The simulation comparison test results show that the improved model is effective and efficient.