The network log file database is important to security management. But, with the development of Internet, it has become too bulkiness to be used directly. In this paper, the wavelet decomposition is adopted to mining the useful rules from the network security data. This is different from traditional data mining methods of association rules. Multi-resolution analysis of wavelet transform method is used to analyze and mine the data of network security log file database on different time scale. The period law of the attack number in different time domain is found by this method. These are important messages for taking relevant measures to insure network security.