【作者】 赵鹏；

【导师】 王苹；

【作者基本信息】 哈尔滨工业大学 ， 船舶与海洋工程， 2025， 硕士

【摘要】 增材制造技术以其高设计自由度和高生产灵活性,在复杂金属构件的成形制造中展现出显著优势,进而在多个领域得到了广泛应用。然而,由于其独特的制造过程,增材成形构件内部不可避免地会存在诸如气孔、未熔合等缺陷。这些缺陷的存在对构件的疲劳性能造成了严重影响,且缺陷的分布和形貌特征具有显著的随机性,使得其对疲劳性能的影响程度难以进行精确量化。这一现状给增材成形构件的服役安全性评估带来了巨大挑战。然而,随着机器学习技术的不断进步,为这一问题提供了有效解决方案。本文选取电弧熔丝增材成形（Wire and Arc Additive Manufacturing,WAAM）TA3钛合金作为研究目标,对其进行了深入的微观组织观察及力学性能测试。测试结果表明,缺陷的存在显著增大了疲劳试验数据的离散性。为了量化评估缺陷对增材成形构件疲劳性能的具体影响,本文引入了缺陷敏感性的概念,并详细阐述了三种敏感性评估指标,即缺陷尺寸（area）^1/2、应力集中系数K_t以及应力强度因子K,本文针对缺陷K_t的准确评估展开研究。鉴于缺陷特征与整体性能之间的紧密联系,采用CT扫描成像技术对WAAM成形TA3钛合金的内部缺陷进行精准检测。基于所获取的检测结果,对缺陷的尺寸、位置及形貌这三大关键特征进行统计分析。以统计结果作为建模依据,开展缺陷K_t的有限元仿真分析,通过调节缺陷模型特征参数来探究K_t的变化规律,结果表明缺陷K_t由缺陷位置参数D/a和形貌参数a/c共同决定。将609组有限元仿真数据划分为训练集和测试集,基于训练集数据和支持向量机回归（Support Vector Regression,SVR）方法,构建了基于缺陷特征的K_t预测模型,此模型对K_t的预测具有较高准确性。根据构建的SVR模型和疲劳试样断口SEM图对各裂纹源缺陷的敏感性评估指标数值进行计算,并利用疲劳试验数据对各指标的准确程度进行验证。在传统指标准确程度不高的基础上,本文提出新指标K^*,,其与疲劳命在在对对数标系下的线性拟合优度达到0.87。通过对疲劳试验数据进行回归拟合,得到WAAM成形TA3钛合金的命在预测K^*-N曲线和S-N曲线。在实际应用时,可根据构件内部缺陷的敏感性高低来选择合适的曲线进行命在预测。总体来说,本文综合运用力学表征试验、有限元仿真以及机器学习技术,提出了一套针对增材制造缺陷敏感性的精确评估指标预测方法。在此基础上,进一步构建了基于缺陷敏感性的WAAM成形TA3钛合金疲劳命在预测模型,为有效评估增材成形钛合金构件的服役安全性提供了切实可行的新途径。更多还原

【Abstract】 Additive manufacturing technology has significant advantages in the shaping and manufacturing of complex metal components due to its high design freedom and production flexibility,thus it has been widely applied in various fields.However,the unique manufacturing process inevitably results in defects inside additive-formed components,such as porosity and lack of fusion.Studies have shown that the presence of defects seriously affects the fatigue performance of additive-formed components.However,due to the random distribution and diverse morphology of defects,the degree of harm to fatigue performance is difficult to quantify,posing great challenges to the safety assessment of additive-formed components.The development of machine learning technology provides a new approach to solve this problem.This paper takes Wire and Arc Additive Manufacturing(WAAM)TA3 titanium alloy as the research object,and conducts in-depth microscopic observation of its microstructure and mechanical performance testing.The test results indicate that the presence of defects significantly increases the dispersion of fatigue test data.In order to quantitatively evaluate the specific impact of defects on the fatigue performance of additively manufactured components,this paper introduces the concept of defect sensitivity and elaborates on three sensitivity evaluation indicators,namely defect size,stress concentration factorK_t,and stress intensity factor K.This study conducts research on the accurate evaluation of the factorK_t.Considering the close connection between defect characteristics and overall performance,CT scanning imaging technology is used for precise detection of internal defects in WAAM TA3 titanium alloy.Based on the obtained detection results,statistical analysis is conducted on three key features of the defect:size,position,and morphology.Using statistical results as the basis for modeling,the finite element simulation analysis of defectK_t was conducted to explore the variation law ofK_tby adjusting the characteristic parameters of the defect model.The results indicate that theK_t of internal defect in WAAM TA3 titanium alloy is jointly determined by the defect position parameter D/a and the morphology parameter a/c.The 609 groups of finite element simulation data were divided into a training set and a test set.Based on the training set data and the Support Vector Regression(SVR)method,aK_tprediction model based on defect characteristics was built,which has high accuracy in predictingK_t.Based on the SVR model and SEM images of fatigue specimen fracture surfaces,the sensitivity evaluation index values for each crack source defect were calculated.The accuracy of each index was verified using fatigue test data.Based on the low accuracy of traditional indicators,this paper proposes a new indicatorK~*,whose linear fit with fatigue life under a double logarithmic coordinate system reaches 0.87.Through regression fitting of fatigue test data,the life predictionK~*-N curve and S-N curve for WAAM TA3 titanium alloy are obtained.In practical applications,an appropriate curve can be chosen for life prediction based on the sensitivity of internal defects in the component.In summary,this paper comprehensively uses mechanical characterization tests,finite element simulation,and machine learning technology to propose an accurate evaluation indicator prediction method for the sensitivity of additive manufacturing defects.On this basis,a fatigue life prediction model of WAAM formed TA3 titanium alloy based on defect sensitivity is further developed,providing a practical new approach for effectively assessing the service safety of additive manufacturing titanium alloy components.更多还原