根据冲压模具失效产生声发射信号的特点，确定失效信号的特征参数；以小波包分析技术提取模具信号的能量特征向量，把各频带内的能量与总能量之比为特征参数；再对各频带的特征参数进行遗传算法生成新的优化特征参数，以概率一致性原理为基础，通过模糊数学可能性理论得到特征参数的隶属度函数，进而对冲压模具的工作状态进行判定，实验证明，该方法能可靠地对模具进行状态的判别，具有很强的实用性。

According to the characteristics of the acoustic emission signal which was induced by punching die when it fail, and though which to determine the characteristic parameters of failure signal; extracting the energy eigenvector of signal failure die by wavelet packet analysis technology, and Take the comparison between the energy in different frequency bands energy and total energy as the characteristic parameters. then though genetic algorithm to optimal the characteristic parameters. the principle of probabilistic consistency and the fuzzy mathematics possible theory were taken as the foundation to obtain the membership grade function of characteristic parameter .Last, the method was used to identify the state of punching die. Experiment shows that the method can reliably discriminate the conditions of the punching die and it has strong practicability.