Concept lattice was creatively used in mining correlated policy rules in this study. It takes much time for a conflict detection routine to search every policy in policy repository with conventional policy storage models to see if conflict occurs before a new dynamic policy is added to the policy repository. A novel storage model for dynamic policies was proposed to address this problem. Dynamic policies were organized into a concept lattice to be grouped effectively and stably. Then a correlated policy rules algorithm based on the concept lattice was proposed after the concept of correlated policy rules was defined. The algorithm greatly reduces the number of policies necessary for conflict detection, and increases efficiency of conflict detection. Performance analysis and simulation show that the proposed algorithm is effective.