Monitoring, diagnostic and prognostic methods - III
Chair: Prof. Myong Kee JEONG (Rutgers University)
PHM Frontiers: Physics-based Approach to Deep Learning Prof. Byeng D. YOUN, Joon Ha JUNG, Myungyon KIM and Jin Uk KO (Seoul National University)
Journal bearing rotor systems are widely used in various industrial applications. Due to uncertainties, the rotor may not behave as it is expected, and thus accidents may happen, which causes economic losses. To prevent such incidents, diagnosing the health state of rotor systems is of critical importance. Conventional health diagnosis approaches heavily rely on the physical features of the rotor systems, which require large amount of resources to develop. To minimize the amount of time and effort required for the feature engineering process of the rotor systems, the deep learning based feature engineering method is proposed in this paper.
Among various deep learning techniques, convolutional neural networks (CNN), which is known to be powerful in recognizing images, is used. To use CNN, vibration signals are transformed into images using the omni-directional regeneration (ODR) method. The ODR method can generate virtual vibration signals in any direction from the actual vibration signals acquired from the two orthogonally placed proximity sensors. Then, the virtual ODR signals are transformed into images that sketch vibration signals acquired at all directions. Lastly, the image patterns are recognized by CNN, which replaces the expensive feature engineering step used in the conventional methods. The proposed deep learning based feature engineering is validated with the datasets from the journal bearing rotor testbed. It is demonstrated that the deep learning based anomaly detection of the journal bearings can substantially improve the efficiency and accuracy, compared to the conventional anomaly detection methods.
Combustion and Resilient Mounting Condition Diagnostics thru Structural Vibration Monitoring of a Diesel Generator Set Prof. Ronald Dela Cruz BARRO, Don Chool LEE, Seok Man SON and Hee Soo KIM (Mokpo National Maritime University, Korea Electric Power Research Institute(KEPRI))
Diesel engine real time condition can normally be depicted by its operational parameters. Condition management, on the other hand, can keep track of the machines’ status on a long term approach utilizing vibration amplitude and resonance. The mounting support condition along with cylinder combustion pressure excitation influences the diesel engine structural vibration dynamics that may lead to component damage and misalignment. As such, engine mounting condition-based monitoring as a tool can be applied to continuously check on its damping limitations brought by material deterioration.
In this paper, a global vibration measurement in accordance with ISO 13373-1 with the measuring points and direction done in accordance with ISO8528-9 was carried on a large diesel generator set installed in a power plant. The test engine is coupled on a generator and fitted with a torsional damper. Vibration analysis established the occurrence of engine structural vibration phenomena at lower frequency range (<200Hz). It was also confirmed that the actual state of the resilient mount can be verified by the structural vibration natural frequency of the engine.
Resilient-Based Control Reconfiguration of Autonomous Systems Mr. Sehwan OH, Benjamin LEE, Michael BALCHANOS, Dimitri MAVRIS and George J. VACHTSEVANOS (Georgia Institute of Technology)
This paper introduces a design methodology for resilient-based control reconfiguration strategy of unmanned autonomous systems (UAS). UAS are actively utilized in the commercial and military arenas, and expected to be deeply immersed in our lives in the near future. One of the critical hurdles of UAS deployment is, however, safety assurance in the presence of ‘incipient failures.’ This concern is highlighted in the recent announcement of regulation proposals for commercial unmanned aerial vehicles (UAV) from the US Federal Aviation Administration in 2015. The proposal states that operations of small UAVs under 55 pounds are limited to ‘line-of-sight’ visibility and daylight-only operations, which will greatly restrict the potential usefulness of civilian/commercial UAVs. It clearly shows that improving and guaranteeing UAS safe operations is a cornerstone for successful fulfillment of UAS utilization. This paper employs the concept of resilience as an advanced fault tolerant strategy. Resilience was defined by Hollnagel as the intrinsic ability of a system to adjust its functioning prior to, during, or following changes and severe disturbances, so that it can sustain required operations even after a major mishap or in the presence of continuous stress. This concept requires major capabilities: detection, prediction, planning, and action. These capabilities can be achieved by an intelligent software core as well as a proper hardware complement. This study contributes to a general design framework of 1) evaluating fault effects on entire system capability by particle filtering-based online fault diagnostics and prognostics, and 2) based on the prediction, reconfiguring control authorities such as control references and mission profile by nonlinear model predictive control and machine learning approaches as a trade-off strategy between system performance and system remaining useful life. The proposed methodology is applied and tested using an autonomous and ground-operable hovercraft as a scaled test-bed. The Aerospace Systems Design Laboratory at the Georgia Institute of Technology built such a test-bed hardware, and the corresponding system and sub-system dynamics models have been developed in Robot Operating Systems (ROS) and Gazebo 3D simulator for the purpose of fast simulations and verification. The hovercraft is used to demonstrate the efficacy of the proposed control reconfiguration framework.