Process control of laser metal deposition

Liu, Y 2018, Process control of laser metal deposition, Doctor of Philosophy (PhD), Engineering, RMIT University.


Document type: Thesis
Collection: Theses

Attached Files
Name Description MIMEType Size
Liu.pdf Thesis application/pdf 11.01MB
Title Process control of laser metal deposition
Author(s) Liu, Y
Year 2018
Abstract Laser metal deposition (LMD) is a novel manufacturing technology that is capable of explicitly fabricating metallurgic products with complex 3D structure, and it has been widely and successfully applied in a variety of industries. However, due to the LMD process is very sensitive to fluctuations in the processing conditions. The geometrical quality of LMD's products may be low or inconsistent. Hence, the overarching research objective of this thesis has been to investigate and propose an approach to stabilize the process by controlling the melt pool at a constant size. This objective was broken down into two major elements: first, how to measure the melt pool size accurately; and second, how to control the melt pool size. To commence the research, a process monitoring system was built in order to measure the melt pool size. The monitoring system consisted of a near-infrared monochrome camera, a narrow band pass filter and a real-time melt pool image process algorithm. Since the grey image was the only format of the camera output, a thresholding based melt pool size measuring algorithm was developed and calibrated. Further, the measuring algorithm was developed to find melt pool size as accurate as possible. Then the algorithm was optimized and accelerated to ensure the measurement can be finished within the sampling interval. In tandem with the development of the monitoring system, a linear parameter-varying (LPV) model that depicts the relationship between laser power and melt pool size (LP-MPS) was estimated. Few deposition experiments with different laser power showed that a first-order transfer function with time delay is the form of the LPV model whose steady state gain and time constant vary according to the process condition. Thus process condition and process condition indices were defined to index the parameters of the LPV model by process conditions. Following this, a PI controller and an MPC were designed to control the melt pool size, and these controllers were tested under a variety of process settings. The stable region of the PI parameters was analysed by Routh-Hurwitz stability criterion. Next, the PI parameters were found for a specific process condition. However, the performance of the PI controlled system would drop significantly when the reference was changed which indicates that each group PI parameters can only be applied to specific process settings. Hence, as an improvement of the PI controller in terms of better dynamic performance and wider suitable process condition range, an MPC was designed based on the state space model which is a non-minimum state space realisation of the LPV model when considering a non-zero reference. Due to the melt pool size measurement algorithm has already occupied a large share of limit computation power, the MPC gain as a constant which was found offline rather than solving the quadratic problem online to find optimised controller output. The test results showed that the MPC could deliver better dynamic performance than the PI controller if the initial voltage was limited precisely or the activation of the laser the MPC can be synchronised. Also, a wider suitable process condition range of a specific MPC gain was observed. Finally, a small brick was built with MPC controlled which further proves that MPC can keep the melt pool size at the set point all the time and reject disturbances quite fast.
Degree Doctor of Philosophy (PhD)
Institution RMIT University
School, Department or Centre Engineering
Subjects Manufacturing Processes and Technologies (excl. Textiles)
Control Systems, Robotics and Automation
Keyword(s) Process Control
Laser Additive Manufacturing
Melt Pool Size
PI control
Model Predictive Control
Versions
Version Filter Type
Access Statistics: 109 Abstract Views, 56 File Downloads  -  Detailed Statistics
Created: Tue, 02 Oct 2018, 10:24:25 EST by Adam Rivett
© 2014 RMIT Research Repository • Powered by Fez SoftwareContact us