动态运行工况下多维失效相关系统可靠性建模及维护决策研究

Reliability modeling and preventive maintenance of systems are crucial for safe production and cost control. Modern production systems become increasingly complex and flexible. Systems may consist of failure-dependent components and work in time-varying operating condition. Few research has considered the reliability modeling of systems subject to the joint effect of dependent failure process, operating condition and maintenance. The project will develop the reliability model and a condition-based preventive maintenance policy of systems subject to multiple dependent failure processes in time-varying operating condition, considering the influence of operating condition on the reliability of components with different failure models. Firstly, the reliability model of systems working under time-varying operating condition will be developed based on the reliability evolution models of components with different failure models. Then, the reliability model of systems after maintenance will be presented based on the maintenance models of components in time-varying operating condition. Finally, a condition-based preventive maintenance policy based on current operating condition will be proposed to determine the preventive maintenance component group dynamically along with the update of operating condition. The project aims to improve the ability of the reliability assessment and maintenance decision-making of complex systems.

系统的可靠性建模以及预防维护决策对于企业安全生产、降低运营成本意义重大。现代生产系统趋向于复杂化及柔性化，系统部件之间存在着失效相关性，同时可能运行在动态的工况下。少有研究考虑系统在失效相关、动态工况、维护联合作用下的可靠性演化规则。本项目从分析工况对不同失效模式部件的可靠性演化影响出发，建立动态工况下系统执行维护后的可靠性模型，并提出基于当前运行工况的视情维护策略。首先剖析不同失效模式部件在动态工况下可靠性演化规则，考虑工况对失效相关作用的影响，建立系统在动态工况下的可靠性模型；继而构建部件在动态工况下的维护效果模型，建立系统在动态工况下的维护可靠性模型；最后考虑未来工况不可预知，以系统长周期维护成本率为目标函数，建立基于当前工况的视情维护策略，随着工况的变化动态地决策出维护部件集以及阈值。本项目旨在提高复杂系统在动态工况下的可靠性预测准确度，增加维护决策的科学性。

科学的预防维护计划对于支撑企业安全生产、降低运营成本具有重要意义。项目考虑现代生产系统面临的运行工况时变性、构成部件相关性、维修样式多样性，开展动态运行工况下多维失效相关系统可靠性建模及维护决策研究。取得以下成果：1）建立不同失效模式部件在动态运行工况下的可靠性演化模型及参数估计方法，并应用于公开数据集，结果证明了模型与参数估计方法的有效性；2）提出面向多维失效相关部件的视情维护策略，应用于液压滑阀等系统，结果证明该策略较传统策略能有效地降低长周期维护成本率；3）提出面向多部件系统动态运行工况下的预防维护策略，考虑部件之间的相关性与维修样式多样性，通过算例证明该策略的有效性。除了以上项目申报时提到的研究内容，我们基于上述研究成果对项目进行了一定的扩展和探索，1）提出多维数据驱动的剩余有效寿命预测方法，应用于铅酸电池、数字通讯系统等，结果证明预测精度较已有方法有明显的提升；2）面向两机流水线系统，考虑运行工况随着作业变动，构建预防维护与作业排序联合决策模型，通过算例证明了模型的有效性。项目的研究成果将丰富复杂系统预防维护决策理论研究，推动复杂系统在动态运行工况下的可靠性建模精度提升，以及有效地增加维护决策的科学性。