针对大多可靠性工程问题中机构极限状态函数为隐式的情况,提出了一种基于极限学习机(ELM)回归近似极限状态方程的可靠性及灵敏度分析的新方法.通过极限学习机与蒙特卡洛法相结合,利用极限学习机快速学习的能力,将复杂或隐式极限状态方程近似等价为显式极限状态方程,运用蒙特卡洛法计算出机构的失效概率,然后由高精度的显式极限状态方程进行各随机变量参数的灵敏度分析.该方法采用了基于单隐层前馈神经网络极限学习算法,因而在拟合非线性极限状态方程上表现优越,计算精度和效率高.最后以某型起落架中可折支撑锁机构为对象,进行了机构的可靠性及敏感度分析.结果表明:该方法具有高精度和高效率的优点,在工程应用上具有一定的价值.%For the most reliability problems with implicit limit state function in engineering mechanism, a new method based on extreme learning machine (ELM) that regress approximation of the limit state equation for reliability and reliability sensitivity analysis was presented. Using the fast learning capacity of extreme learning machine, implicit limit state function was approximated to explicit limit state function by combining the extreme learning machine with Monte Carlo method. The failure probability of mechanism was calculated using Monte Carlo method, and then explicit limit state function could be employed to analyze the sensitivity of each parameter. The method using single hidden layer feedforward neural network learning algorithm, thus in fitting the nonlinear limit state equation performed superior and has a high accuracy and efficiency. Finally, the reliability and sensitivity analysis of the folding support lock mechanism in a type of landing gear were carried out using the presented method. The results show that the presented method has virtues of high accuracy and efficiency. Thus, it is valuable for the engineering application.
展开▼
