涡轮盘概率风险的评估方法

针对CCAR33.70条款对涡轮盘开展概率风险评估的要求，研究了响应面法在定义重点采样区间以及计算裂纹扩展寿命中的应用，给出了在蒙特卡洛模拟过程中应用响应面法定义失效域，缩短样本方差、提高抽样效率的方法.分析结果表明应用重点采样法可以极大地提高抽样效率.采用重点采样法获得的失效样本数与总样本数的比例高达94%，远大于传统的直接抽样法获得的失效样本数与总样本数的比例（仅为33%）.该方法可以在保证分析精度的条件下减少样本的数量.另外，应用响应面法拟合寿命近似方程，可以极大地缩减裂纹扩展寿命的计算时间:利用寿命近似方程开展104次随机抽样所花费的时间仅有2min，远低于应用传统寿命计算方法所花费的时间（288 min）.通过对寿命近似方程的寿命计算误差分析可以看出:最大计算误差仅有1.0%，计算结果与传统裂纹扩展寿命模型的计算结果吻合度高.该研究成果可以提高涡轮盘概率风险评估的效率. 

Probabilistic risk assessment method for turbine disk

For the turbine disk probabilistic risk assessment requirement of CCAR 33.70 airworthiness regulation, the applications of response surface method used in definition of failure zone and estimation of crack growth life were studied. A method of variance reduction of sampling and improving sampling efficiency was demonstrated based on the response surface method to allocate the failure sampling zone during Monte Carlo simulation. It was shown that importance sampling method could significantly improve sampling efficiency:samples rate between failure samples and total samples was 94% by importance sampling method, bigger than traditional direct sampling method (only 33%). The importance sampling method can minimize the total number of samples required to achieve a desired analysis accuracy results. In addition, the life approximation function by response surface method was presented to significantly reduce the computation time of crack growth life:computation time for 10⁴ samples was only 2 minutes by using life approximation function, which was much shorter than traditional life model (288 min). The life error analysis showed that the maximum error was only 1.0%, which had good agreement with traditional method of crack growth life. These results can be applied to improve the efficiency of turbine disk probabilistic risk assessment.