Abstract:

The fuzzy ISODATA （Iterative Self-Organization Data）clustering method is adopted for classification of individual samples so as to generate the cluster centers immediately, and consequently the samples to be recognized can be identified. The selection principle of initialization matrix U_fo for this method is also discussed. The fuzzy ISODATA method is compared with the fuzzy equivalent matrix method in recognization and calculation of metal cutting states. The results prove that both fuzzy pattern recognition methods are quite applicable to recognition of the metal cutting states, and they can improve the accuracy of pattern recognition and simplify the calculation, so that they are advantageous to realizing the real time recognition and adaptive control of the metal cutting states.