跳跃跟踪SSA交叉迭代AP聚类算法OA北大核心CSTPCD

Aiming at the problem that the traditional affinity propagation(AP)clustering algorithm takes the similari-ty between data points as the input measure,and the accuracy of the algorithm cannot be accurately controlled due to the need to preset the preference(p)and the damping coefficient(λ),a jump tracking sparrow search algorithm(JTSSA)opti-mized hybrid iterative AP clustering method(JTSSA-AP)is proposed.Firstly,in order to solve the problem of insufficient update of the position of the producers and the scroungers in sparrow search algorithm(SSA),a jump tracking optimization strategy is designed.By considering the preference factor,the jump strategy updates the producers in a large step,which in-creases the global exploration ability and optimization speed of SSA algorithm.The scroungers dynamically track the up-date position of the leading sparrow in a small step,and uses the adaptive population division mechanism to update the pro-portion of the producers and the scroungers,which increases the late local development ability and optimization speed of the algorithm.Secondly,on the basis of the original Tent mapping,the disturbance factor is added and three parameters are added,so that the mapping distribution range is increased and the small periodic point and unstable periodic point are avoid-ed.Finally,the silhouette index is introduced as the evaluation function,JTSSA is designed to automatically find better p and λ instead of manual input parameters,the tent map based on disturbance factor optimize the AP clustering,and the opti-mal number of clusters is determined by hybrid iteration.Multiple algorithms are utilized to cluster the 10 public datasets of the university of California Irvine dataset.Simulation results indicate that the proposed clustering algorithm in this paper ex-hibits superior search efficiency and clustering accuracy compared to the AP algorithm,the AP clustering algorithm based on differential evolution,the AP clustering algorithm optimized by SSA,and the evolutionary affinity propagation.Cluster analysis is conducted on country data,and the proposed method demonstrates greater accuracy,effectiveness,and rationali-ty,showcasing considerable practical value.