In order to effectively improve the safety level of tower crane in China and reveal the internal relationship of tower crane accidents, the improved Apriori algorithm was used to mine the association rules of tower crane accidents. Firstly, 210 accident investigation reports of tower crane were collected. According to the accident investigation report and literature analysis, the basic attributes and cause attributes of tower crane accidents were extracted. Then, based on different accident stages, an improved association rule algorithm based on combination weighting and directional constraints was proposed, and the influencing factors of tower crane operation accidents were quantitatively analyzed to mine the potential mapping relationship. Finally, according to the mining results, the key association rules of tower crane accidents were analyzed to provide theoretical guidance for the improvement of the safety level of tower crane accidents, and relevant suggestions are put forward.