The invention discloses a clustering center determining method and a clustering center determining
system. The method includes the following steps that: a water
wave group including a plurality of water
waves is constructed; propagation
processing is performed on each water wave in the water
wave group; and whether the fitness value of each water wave obtained after the propagation
processing is performed is greater than the fitness value of the corresponding water wave obtained before the propagation
processing is performed judged; if the fitness value of each water wave obtained after the propagation processing is performed is greater than the fitness value of the corresponding water wave obtained before the propagation processing is performed, the water
waves obtained after the propagation processing are adopted to replace the corresponding water
waves which are obtained before the propagation processing; if the fitness value of each water wave obtained after the propagation processing is performed is not greater than the fitness value of the corresponding water wave obtained before the propagation processing, the
wave height and wave length of the water waves in the water
wave group which are obtained before the propagation processing are updated, and the locations of the water waves which are obtained before the propagation processing are kept unchanged; the wave group and the number of iterations are updated; whether the current number of iterations is smaller than a set
evolution algebra is judged; if the current number of iterations is smaller than the set
evolution algebra, the method shifts to previous steps, propagation processing is performed on the water waves in the updated water wave group; and if the current number of iterations is not smaller than the set
evolution algebra, water waves in the updated water wave group, which have the largest fitness value, are selected as optimal water waves. According to the method and
system provided by the invention, from the perspective of optimization search, a clustering center is optimized from generation to generation by means of propagation processing, and the clustering center can be approximate to an optimal clustering center.