Equipment fault prognosis is significant for safeguarding the safe operation of equipment and improving the efficiency of equipment management. Temporal fault data model was built by using Apriori traditional association rules algorithm based on the characteristics of fault data. Improved Apriori algorithm and frequent temporal association rules algorithm were proposed by converting fault data to temporal item sets matrix. Equipment fault trends were predicted by mining the frequent temporal association rules of fault data based on the algorithm, which provided strong support for equipment management. At last an example was given to prove the feasibility.