CAJ | 학술논문

针对模糊C-均值FCM（Fuzzy C-Means）聚类算法易陷入局部最优解，对初始值敏感的缺点。提出基于混沌和动态变异蛙跳SFLA（shuffled frog leaping algorithm）的FCM算法。该算法先用混沌的Tent序列初始化青蛙群体以增强群体的多样性，提高初始解的质量；并根据青蛙的适应度方差值选择相应的变异概率。再将改进后的蛙跳算法优化FCM算法，最后求取全局最优。人工数据及经典数据集的仿真结果表明，该算法（CMSFLA-FCM）与SMSFLA-FCM、SFLA-FCM和FCM聚类算法相比，寻优能力更强，聚类效果更优。
침대모호C-균치FCM（Fuzzy C-Means）취류산법역함입국부최우해，대초시치민감적결점。제출기우혼돈화동태변이와도SFLA（shuffled frog leaping algorithm）적FCM산법。해산법선용혼돈적Tent서렬초시화청와군체이증강군체적다양성，제고초시해적질량；병근거청와적괄응도방차치선택상응적변이개솔。재장개진후적와도산법우화FCM산법，최후구취전국최우。인공수거급경전수거집적방진결과표명，해산법（CMSFLA-FCM）여SMSFLA-FCM、SFLA-FCM화FCM취류산법상비，심우능력경강，취류효과경우。
Fuzzy C-means(FCM)clustering algorithm is prone to fall into the solution of local minimum and is sensitive to initial value. Aiming at these drawbacks,we present a fuzzy C-means clustering algorithm which is based on chaotic and dynamic mutation shuffled frog leaping algorithm (SFLA ).In this algorithm,frog population is initialised with Tent chaotic sequence to enhance the diversity of the population and to improve the quality of initial solution,and the corresponding mutation probability is selected according to the fitness variance of each frog.Then the improved shuffled frog leaping algorithm is employed to optimise the FCM algorithm,and to get the global optimum finally.Simulation results on artificial data and classic dataset show that compared with the SMSFLA-FCM,SFLA-FCM and FCM clustering algorithms,the new algorithm (CMSFLA-FCM)has stronger optimisation ability and better clustering effect.