一、报告人
徐正国博士（浙江大学副教授）
二、摘要
Abstract—An accurate prediction of Remaining useful life (RUL) is highly dependent on an appropriate characterization of the degradation process. In operation, a device may suffer from both defects and shocks, such as the environmental variations and various loads, which leads to a non-stationary degradation process. Thus, a Wiener-process-based adaptive hybrid degradation model is proposed to characterize the degradation process with shocks. A filtering algorithm based on Interacting Multiple Models (IMM) is developed to estimate the system state, followed by an expectation-maximization (EM) algorithm and an IMM filtering-based smoother to estimate the unknown parameters. The proposed methods can achieve an accurate estimation by taking into account all the historical condition monitored (CM) measurements. In addition, a closed-form expression of the RUL distribution is achieved. Finally, a practical case study is presented to illustrate the proposed approach.
三、地点与时间
机械工程学院，机械楼D518，2019年7月15日 8:30-10:00
四、报告人简介
徐正国，浙江大学控制科学与工程学院副教授，博士毕业于清华大学自动化系；2009年3月至今，在浙江大学控制科学与工程学院工作。主要研究方向为工业大数据分析、复杂工程系统可靠性分析和故障诊断、基于人工智能理论的工程优化与诊断技术。已获得国家科技进步一等奖1项、教育部自然科学一等奖1项、中国自动化学会自然科学一等奖1项；承担国家自然科学基金、国家863计划项目、国家科技支撑计划项目等各类项目近20项；在国内外期刊及会议发表相关论文40余篇；获得授权国家发明专利8项。现任中国自动化学会技术过程的故障诊断与安全性专业委员会委员、中国系统工程学会系统可靠性工程分会理事。
