W288 the development of automation, the dispatching automation system need to handle a large amount of messy signals. It is significant to ensure system stability by recognizing and classifying signals quickly and accurately. This paper introduces a learning algorithm based on the feature weight of CFuzziness function, which could solve dimension disaster problem of fuzzy C mean algorithm. By minimizing the function CFuzziness with the gradient descent algorithm, a proper weight value can be given to each feature, and the weighted fuzzy C mean algorithm is obtained. The results show that the clustering results of weighted fuzzy C mean algorithm are superior to the fuzzy C mean algorithm.