Characterization and Control of Non-Steady State Machine Vibration

点击翻译按钮获取中文摘要

Non-steady state operating conditions are commonplace in a wide range of machinery from wind turbines to airplanes. However the system response dynamics in these situations remain poorly characterized and not well understood. The changing operating conditions that generate non-steady state responses are due to a variety of factors such as unpredictable machine speed (planetary gear systems in wind turbines), fluctuating mass (loading and emptying of mining skips) or variable excitation (aircraft flight envelopes, MRI scanner imaging excitation). In all cases it is extremely important to fully understand the system response in order to control mechanical vibration and mitigate potentially harmful effects.The broad class of non-steady state machine response applications has a wide and variable time scale (potentially ranging from minutes to months) defining the different operational regimes. The many different parameters that define the operating conditions of these systems may also combine in unpredictable ways. In these cases it remains challenging to characterize, predict and control the mechanical vibration.Non-steady state system characterization has followed an ad-hoc development path to date. Most work has focused on particular applications where the non-steady state system behaviour is unique. A more structured approach will be followed in the proposed research work. This approach will begin with a set of characterization criteria that are traditionally used in steady state system identification and characterization and develop new and/or hybrid methods for characterization of non-steady state operation. Preliminary work by myself and others has revealed several new signal analysis methods that show significant promise in this regard.Recent research has begun to produce tools that allow system condition monitoring, fault diagnosis, vibration control and system performance optimization. Most of this work requires fixed time scales, prior knowledge of the expected system operating conditions and a reasonable history of sample data representing those operating conditions. My past research work has been directly related to this research area and was focused on methods that do not require detailed prior system knowledge nor sample data for training of analysis algorithms. Other closely related work that I have completed recently includes noise and vibration reduction in MRI scanners and vibration control in mining skips. Solutions to the vibration and noise problems in both of these applications relied on development of a clearer understanding of the system performance during non-steady state operational situations.The research work proposed in my Discovery Grant application will be focused on the further understanding of non-steady state system performance through system characterization. Application of the results of this work will lead directly to new methods for machine/structure condition monitoring, fault detection and diagnosis, vibration control, and system performance and design optimization. In my research program I will be focusing on the application of these results as they apply to fault detection and diagnosis in planetary gearboxes, noise reduction in MRI scanners and vibration control in mining skips. While these applications may seem relatively specific they have common aspects and represent a broad field of study of interest to a range of important resource based industries in Canada. My expertise and experience uniquely position me to address these topics.