性能数据驱动的滚动轴承可靠性演变过程预测方法

With the devolopment of the fields of aeronautics and astronautics, bullet trains, and alternative energy, reliability research of rolling bearings has attracted much attention. It is most pressing required that information on rolling bearing performance reliability at the future time with a timely forecast will be useful for the early detection of the hidden danger of failure of rolling bearing performance, thus avoiding serious accident. Degradation of much performance of rolling bearings is uncertain along with unknown evolvement of reliability due to unknown trend and failure probability distribution in advance and complexity and polytrope of intrinsic and external factors, resulting in this issue unresolved for the existing reliability theory. For this end, this project studies the forecasting method for performance-data-driven reliability evolvement of rolling bearings based on the theory of chaos, mechanics, information entropy, fuzzy set,and Bayesian statistics. A concept of performance-data-driven failure processes is proposed and the forecasting model of performance evolvement from no failure to failure is established to reveal the mechanism of performance failure evolvement; a fusion self-healing evolution model is presented to make a representation of the reliability evolvement trajectory; a concept of double-driving statistics is defined to depict new connotations of hypothesis testing rejection region of reliability evolvement; a concept of characteristic quantity of joint evolution is developed to explore dynamics nature of reliability evolvement probability. The research findings can move beyond the limitations of classical reliability theories, make an advance in understanding the unknown features of rolling bearings, provide new idea for deploitation of the frontiers of performance reliability evolution theories, and deepen basic science research for reliability.

随着航空航天、高速客车与新能源等领域的发展，滚动轴承可靠性研究备受关注，迫切需求及时预测其未来性能可靠性信息，尽早发现失效隐患，避免恶性事故发生。因趋势与失效概率分布事先未知，且内、外部因素复杂多变，使滚动轴承很多性能的退化具有不确定性并伴随可靠性的未知演变，现有可靠性理论难以解决这个问题。为此，基于混沌、力学、信息熵、模糊集合与贝叶斯统计等理论，研究性能数据驱动的滚动轴承可靠性演变过程预测方法。提出性能数据驱动失效过程概念，建立从无失效到失效的性能演变过程预测模型，以揭示性能失效演变机制；提出融合自愈演变模式，以表征性能可靠性演变轨迹；提出双驱统计量概念，以刻画可靠性演变假设检验否定域的新内涵；提出联合演变特征量概念，以探索性能可靠性演变概率的动力学本质。研究成果能突破传统可靠性理论的局限，促进对滚动轴承未知特性认识，为开拓性能可靠性演变研究新领域提供新理念，深化可靠性基础科学研究。

滚动轴承性能可靠性，是指在给定的环境、条件与时间内，滚动轴承运行性能可以满足规定运行性能的能力。这种能力可以用可靠性函数量化表征。若在使用过程中滚动轴承性能不能满足规定的运行性能要求，则认为滚动轴承性能失效。滚动轴承性能退化通常经历初期退化、渐进退化、快速退化与急剧退化等阶段，性能失效轨迹与概率分布、性能可靠性函数等信息随之变化。这样，依赖于已知概率分布与趋势等先验信息的可靠性理论遭遇严重挑战。为此，本项目提出了一种性能数据驱动的滚动轴承可靠性演变过程预测方法，在性能退化趋势与失效概率分布等先验信息未知条件下，仅凭当前的无失效性能数据，就可以揭示出未来滚动轴承性能失效及其可靠性演变过程的特征。本项目研究了无失效性能数据的演变特征，在滚动轴承性能无失效期间，可以预测出未来的性能失效时间即失效数据，发现了性能从无失效到失效的内在演变机制；用多种数学概念与思想，研究滚动轴承性能的无失效数据，挖掘出拉格朗日乘子的多个侧面信息，实施了多侧面信息融合，得到了可靠性的融合自愈演变模式；提出了性能可靠性演变过程的双驱统计量概念，研究了等价关系发生概率函数与后验发生概率函数的时变交集，构建了动态的假设检验否定域，表征出遗传与变异对演变性质的驱动功效；提出了联合演变特征量概念，发现了可靠性的演变的力学与随机特征以及多种关系，揭示出滚动轴承性能可靠性演变概率的随机力学本质。科学意义是：性能数据驱动滚动轴承失效过程概念，可以揭示出性能失效的内在演变机制，促进了滚动轴承动态可靠性理论发展；提出的性能可靠性的融合自愈演变模式，可以表征出滚动轴承性能可靠性演变轨迹；提出的双驱统计量概念，可以刻画出可靠性演变假设检验否定域的新的科学意义与内涵；提出的联合演变特征量概念与发现的性能可靠性演变概率的动力学本质，可以促进对滚动轴承未知新特性的认识，为滚动轴承可靠性理论的发展奠定新基础。

内容获取失败，请点击重试