Aiming at a certain type of construction machinery, the knowledge point structure tree is constructed, and the Apriori algorithm is used to analyze its knowledge points. Through the pretreatment and discretization of the student score data over the years, the relevance between knowledge points is obtained. According to the relevance confidence of knowledge points, the confidence threshold of topic selection of each assessment module is set, so as to screen out a more scientific and reasonable combination of test papers. Then, through the new test score data, the relevance of knowledge points is further analyzed, and the confidence threshold setting and test paper combination are optimized. The results show that the algorithm can effectively improve the scientificity of course assessment and the rationality of evaluation method.