AerMet100钢拉伸断口形貌仿真  

Simulation of Tension Fracture Surface Morphology of Steel AerMet100

作　　者：张雁祥[1] 闫牧夫[1] 唐丽娜[1] 

机构地区：[1]哈尔滨工业大学材料科学与工程学院,哈尔滨150001

出　　处：《机械工程师》2008年第9期90-92,共3页Mechanical Engineer

基　　金：国家973专题项目;金属精密热加工国家级重点实验室基金项目

摘　　要：研究了时效工艺对AerMet100钢延性断口形貌的影响。基于图像处理技术实现了对试验用钢拉伸断口微观形貌的模式识别,提取韧窝的面积和深度作为特征矢量,并应用自组织映射(SOFM)神经网络将韧窝形貌分成特点不同的6组。基于BP神经网络原理,建立了由时效工艺预报试验用钢各组韧窝的数量的数学模型。根据预报结果通过图像处理过程,实现了不同时效工艺下的AerMet100钢拉伸断口的仿真,模拟结果与试验结果基本吻合。The effect of aging treatment on ductile fracture surface of steel AerMet100 is investigated. Pattern recognition is implemented by digital processing of SEM fractographs： The area and depth of dimples are extracted as feature vectors. Self-organization feature mapping （SOFM） is applied to classify the dimples into six groups. The model for predicting the number of every cluster in a SEM fractograph as a function of aging temperature and time has been established using a back propagation artificial neural network （BP ANN）. According to the numerical prediction results, the ductile fracture surfaces under different aging treatments are simulated. The simulation results agree basically with the experimental ones.

关 键 词：AERMET100钢 时效 断口 模式识别 仿真 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]