基于分形理论从信号中提取盒维数和信息维数作为分类特征,这种特征包含了信号幅度、频率和相位的变化规律,集中了各种调制方式之间的差异信息,在分类的意义上是有效特征,同时,从理论上分析了它们对噪声干扰不敏感的特性.基于这样的特征进行分类器设计非常简单,识别率较高,有较好的工程应用价值. Based on the fractal theory, the box dimension and the information dimension are extracted from the signal as the classification features. This feature includes the variation rule of the signal amplitude, frequency and phase, concentrates the difference information among various modulation schemes, and in the sense of classification Is an effective feature, and at the same time, it is theoretically analyzed that they are insensitive to noise interference.Classifier design based on such features is very simple, high recognition rate, and has good engineering application value.