【摘要】 为确保工业机器人实现可靠性目标，可靠性分配是其制造设计阶段需要完成的任务。根据工业机器人结构复杂、不确定度高、样品少且组成零件间具有失效相关性的特点，提出了一种基于反向传播（Back Propagation, BP）神经网络和毕达哥拉斯模糊数（Pythagorean Fuzzy Number, PFN）的工业机器人系统的可靠性分配方法，使用Copula函数建立系统可靠性模型，将工业机器人的故障分为系统级、子系统级和部件级3个层次，通过使用BP神经网络，将系统可靠性、子系统结构重要度和子系统复杂度作为输入变量，完成系统级到子系统级的可靠性分配；引入PFN对重要程度、环境条件、技术水平、可维修性、费用灵敏度、复杂性6个影响因素进行评分，从而构造分配权重，完成子系统级到部件级的可靠性分配。结果表明，该方法可以实现可靠性目标，并保证了可靠性的增长。更多还原

【Abstract】 In order to ensure the industrial robot realizes the reliability goal, reliability allocation is a task to be accomplished in its manufacturing design stage. According to the characteristics of industrial robots, such as complex structure, high uncertainty, few samples, and failure correlation between component parts, a reliability allocation method for industrial robot systems based on BP neural network and Pythagorean fuzzy numbers was proposed. Using Copula function to establish a system reliability model, the failures of industrial robots were classified into three levels, the system level, the subsystem level, and the component level. By using the back propagation(BP) neural network, the system reliability, subsystem structure importance and subsystem complexity were taken as the input variables to complete the system-tosubsystem reliability allocation. The Pythagoras fuzzy number was introduced to score the influence factors of importance, environmental condition, technical level, maintainability, cost sensitivity and complexity, complete the reliability allocation from the subsystem level to the component level. The results show that the methodology achieves reliability goals and ensures reliability growth.更多还原