基于机器学习的X80管线钢环焊缝热影响区硬度预测模型OA

To investigate the characteristics of the heat-affected zone(HAZ)in X80 pipeline steel girth weld joints,and obtain the nonlinear relationship between thermal cycle data and HAZ hardness,precise predictions of HAZ hardness were conducted.Using X80 pipeline steel as the research subject,thermal simulation tests were performed.Microhardness tests were conducted on the thermal simulation specimens to determine the hardness values and the hardness distribution of the HAZ in full scale girth weld joints.Based on the hardness database of thermal simulation specimens,a multi-index comprehensive evaluation was carried out to optimize the support vector regression algorithm using particle swarm optimization,thereby constructing a hardness prediction model.Furthermore,characteristic data of the thermal cycle were acquired through finite element modeling of the girth weld.Input the hardness prediction model to predict the actual hardness distribution in the heat-affected zone.Results showed that the relative errors in the predicted mean hardness for fully automatic welding and combined automatic welding were 4.70%and 6.01%,respectively.The model demonstrated high accuracy and could provide guidance for optimizing welding processes,ensuring the inherent safety of high-grade pipelines.