关键词: 螺栓拧紧工艺, 拧紧失效, 分类算法, 接触应力, 智能识别

Abstract: The quality of bolted connections directly determines the performance of equipment,and the quality of these connections is greatly influenced by the tightening process. Improper tightening can lead to structural failure,bolt breakage,structural fatigue,and ultimately major safety accidents.Currently,torque wrench judges the failure mode of tightening based on torque size,but this method cannot quickly obtain the stress state of dangerous sections such as the root of the thread during the tightening process. That is,when the stress is about to exceed the tensile strength,it cannot predict and alarm in advance,causing instantaneous torque to be too high and leading to bolt breakage.In response to the above issues,the relationship between bolt torque and stress is established,and an intelligent method for recognizing and assessing bolt tightening failure is proposed using finite element and deep learning classification algorithms. This method not only achieves an intelligent recognition rate of 99 % for bolt tightening failure modes but also can obtain the trend of changes in pretightening force and contact stress during the bolt tightening process,providing technical support for improving the quality of bolt tightening.

Key words: bolt tightening process, tightening failure, classification algorithms, contact stress, intelligent recognition