Intelligent Reliability Assurance Using Dynamics Modeling and Machine Learning

使用动力学建模和机器学习的智能可靠性保证

• 批准号: RGPIN-2021-02900
• 负责人: Zuo, Mingjian
• 金额: $ 5.32万
• 依托单位: University of Alberta
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2021
• 资助国家: 加拿大
• 起止时间: 2021-01-01 至 2022-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=742272
• 关键词: Intelligent; Reliability; Assurance; Using; Dynamics

Today's society depends on safe and reliable operation of colossal engineering systems such as power plants and air, ground, and water transportation systems. Despite technological advancements in reliability assurance methodologies over the past 80 years, accidents and service disruptions affecting millions of people still occur frequently. Because degradation occurs as a device delivers service and inherent uncertainty exists in material properties, degradation behaviors, and environmental conditions, reliability assurance is a challenging problem to solve. Growing functionalities, faster operating speeds, and harsher operating environments of today's engineering systems present new challenges to their reliability assurance. New approaches including physics-based computer simulations and machine learning algorithms for online reliability assurance are needed to address these challenges. The new research frontier within the reliability community is to develop unique solutions for separate classes of engineering systems such as rotating equipment and pipelines. My research group has been making significant contributions over the past 30 years to the long-term goal of advancing reliability assurance methodologies for engineering systems. The proposed research over the next five years aims to build upon the expertise of my research group and to utilize computing power and artificial intelligence to improve the existing reliability assurance methodologies for the asset class of rotating equipment. Specifically, we will (1) develop physics-based dynamics models for gear systems which generate vibration responses reflecting the degrading health conditions of their critical components and the varying operating conditions including load and speed and (2) develop data-driven machine learning algorithms for accurate assessment of the hidden health state of gear systems, prediction of their remaining useful lives, and effective maintenance decision making to ensure their reliable and cost-effective operation. We will conduct extensive laboratory experiments on gear systems to provide the needed data to validate the developed dynamics models and the generated machine learning algorithms for various aspects of reliability assurance. The proposed research will generate practical models and intelligent algorithms which can be directly applied to systems such as machine tools, wind farms, driverless vehicles, and power plants. These cutting-edge research results together with the trained highly qualified personnel will put Canadian research community and relevant industry at the forefront in the world stage.

当今社会依赖于大型工程系统的安全可靠运行，如发电厂和空中，地面和水上运输系统。尽管过去80年来可靠性保证方法取得了技术进步，但影响数百万人的事故和服务中断仍然频繁发生。由于退化发生在设备提供服务时，并且材料特性、退化行为和环境条件存在固有的不确定性，因此可靠性保证是一个需要解决的具有挑战性的问题。当今工程系统的功能不断增加、运行速度越来越快、运行环境越来越恶劣，这对系统的可靠性保证提出了新的挑战。需要新的方法来应对这些挑战，包括基于物理的计算机模拟和在线可靠性保证的机器学习算法。可靠性领域的新研究前沿是为不同类别的工程系统（如旋转设备和管道）开发独特的解决方案。 在过去的30年里，我的研究小组一直在为推进工程系统可靠性保证方法的长期目标做出重大贡献。未来五年的拟议研究旨在利用我的研究小组的专业知识，利用计算能力和人工智能来改进旋转设备资产类别的现有可靠性保证方法。具体而言，我们将（1）为齿轮系统开发基于物理的动力学模型，这些模型产生的振动响应反映了其关键部件的退化健康状况以及包括负载和速度在内的各种运行条件，（2）开发数据驱动的机器学习算法，用于准确评估齿轮系统的隐藏健康状态，预测其剩余使用寿命，和有效的维护决策，以确保其可靠和具有成本效益的运行。我们将对齿轮系统进行广泛的实验室实验，以提供所需的数据来验证开发的动力学模型和生成的机器学习算法，以确保可靠性的各个方面。这项研究将产生实用的模型和智能算法，可直接应用于机床、风电场、无人驾驶汽车和发电厂等系统。这些尖端的研究成果加上训练有素的高素质人才，将使加拿大研究界和相关行业在世界舞台上处于领先地位。