基于不完全数据的Q345大型角钢疲劳性能研究

为利用疲劳试验中出现的部分不完整疲劳试验数据,将不完全数据整体估计法与单点-似然法相结合建立兼顾效率和精度的概率S-N曲线模型。基于传统单点-似然法,充分考虑不完全疲劳试验数据出现的概率,利用极大似然原理建立带有不完全数据的三参数幂函数S-N曲线表达式回归分析方法,给出概率S-N曲线表达式。同时,对6根规格为∟160×12的大型Q345等边角钢开展疲劳试验研究,观察不同疲劳荷载作用下等边角钢的破坏特征,分析其疲劳寿命的变化规律。利用建议带有不完全数据的概率S-N曲线模型构建方法,建立Q345大型角钢的概率S-N曲线模型。结果表明,与传统单点-似然法的计算值相比,建议带有不完全数据的概率S-N曲线模型构建方法拟合的参数值S₀、m和C分别仅相差1.61%、4.32%和4.15%; Q345大型等边角钢疲劳振动次数越多,缺口越整齐,脆性破坏特征越显著;疲劳寿命随荷载减小而明显增加,荷载由197.82 MPa减小到169.28 MPa时等边角钢的疲劳寿命增加了100.41%;建议方法能有效利用不完整疲劳试验数据建立概率S-N曲线方程。

Fatigue properties of Q345 large equal angles with incomplete test data

In fatigue tests, it is a universal problem that the fatigue life of some specimens cannot be successfully obtained due to the adverse effects of time consuming, high cost, instrument failures and other uncertainties. In order to use these incomplete test data, a new probabilistic S-N curves model, considering both efficiency and accuracy, was established by combining the single-point likelihood method and the integral estimate method using incomplete test data. Based on the traditional single-point likelihood method and the probability of incomplete test data, a regression analysis method was proposed to determine the three-parameter power function of S-N curves by using the maximum likelihood principle. According to the proposed regression analysis method, the probabilistic S-N curve equations were given. Meanwhile, the fatigue tests were conducted on six full-size specimens to investigate the fatigue properties of Q345 large equal angles (∟160×12). According to the experimental results, the effects of different fatigue cyclic loadings on the failure mode and fatigue life of specimens were discussed. At last, the probabilistic S-N curve equations of Q345 large equal angles were then derived using the proposed method. It is found that the proposed method can accurately predict the parameter values of the probabilistic S-N curve equations by comparing the calculated results of the traditional single-point likelihood method. The differences of S0, m and C between the proposed method and the conventional single-point likelihood method are 1.61%, 4.32% and 4.15%, respectively. The brittle failure of specimens becomes more obvious with the increase in cyclic loadings. However, the fatigue life of specimens increases drastically with the decrease in cyclic stress. For instance, the fatigue life is increased by 100.41% when the cyclic stress is reduced from 197.82MPa to 169.28MPa. Furthermore, the proposed method can effectively use incomplete test data to establish the probabilistic S-N curve equations.