项目名称: 数据驱动关键性能指标相关的故障诊断方法研究
项目编号: No.61503039
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 自动化技术、计算机技术
项目作者: 王光
作者单位: 渤海大学
项目金额: 21万元
中文摘要: 复杂工业过程的故障诊断一直是研究的热点问题,基于模型的故障诊断方法由于对系统解析模型的过分依赖而难以在实际系统中推广和应用。为了解决这一问题,数据驱动的故障诊断方法被提出并越来越受到工业界和学术界的的关注。值得注意的是,在过程工业中,系统输出的最终的产品质量往往是人们最关心的关键指标。当系统发生故障时,如果系统输出的产品质量并不受到影响,那么就可以通过降低这类故障的报警率来减少系统停机检修的时间,进而节约维护成本并提高使用效率。因此,研究如何将过程变量空间中的故障进行分类,区分出关键性能指标相关和关键性能指标无关的故障成为实际系统的迫切需求。针对这一实际需求,本课题在复杂工业过程难以建立精确的数学模型的前提下,利用系统传感器获得的历史数据和在线数据,研究线性过程和非线性过程中的关键性能指标相关的故障诊断方法,为确保复杂工业过程的安全和高效运行奠定理论基础。
中文关键词: 数据驱动;故障诊断;质量相关;关键性能指标;复杂工业过程
英文摘要: The fault diagnosis of complex industrial processes is a hot issue. The model-based diagnosis approaches can be hardly used in practice due to their dependencies on the analytical models of systems. To overcome the drawbacks of model-based techniques, data-driven methods were proposed and have been receiving considerable attention in recent years. It is worth noting that the final product quality of industrial process is usually the most important indicator concerned by people. If a fault has no influence on the final product quality, then it is of practical significance to reduce downtime by reducing alarm rates of this kind of fault. Therefore, classifying the faults occurred in process data space into quality-related and quality-unrelated is quite necessary in practice. In view of this, this project aims to develop quality-related fault diagnosis approaches using data-driven techniques to ensure the safety of linear or nonlinear industrial processes.
英文关键词: data-driven;fault diagnosis;quality-related;key performance indicators (KPI);complex industrial process