Abstract: When the secondary system of the intelligent station fails, the association analysis method can be used to mine the historical data to get the fault analysis results. When the traditional association analysis algorithm is applied to process the secondary fault data of the intelligent station, indicators such as support degree and confidence level need to be set to screen the appropriate rules. The setting of indicators is manual, causing the process to be cumbersome. At the same time, a large number of redundant rules will be generated so a lot of manpower will be consumed. Based on the adaptive fireworks algorithm, this paper proposes an association optimization method in which AFWA is combined with ITL mine algorithm. With the help of historical data, the objective function was used to optimize the parameters, which improves the reliability of the rules and ensures that the analysis results cover the whole data. Finally, the rule screening strategy was introduced to set the minimum similarity index. The analysis results were screened to eliminate redundant rules and derivative signals to the greatest extent, so as to improve the readability of rules and ensure that the results are easy to understand. The result verifies that, compared with other traditional optimization algorithms, this method has better convergence performance so that it can save manpower to a certain extent and ensure the accuracy and reliability of the analysis results.