Research Themes
1. Field assisted manufacturing
Complex internal surfaces are required in a variety of applications such as conformal cooling channels in the mould and die industry, waveguides in nuclear power plants, implants in healthcare, etc. Moreover, the rapid development of AM technology has been boosting the production of more complicated parts with unprecedented functionalities. The final surface quality of these non-conventional surfaces greatly affects service performances of the various components such as the cooling efficiency of conformal cooling channels that deteriorate with a poor surface finish. Other examples include energy losses with rough waveguides. Improving the surface quality of non-conventional surfaces is challenging for three reasons from the aspect of manufacturing: (1) poor tool accessibility of conventional finishing tools to reach and conform to these surfaces, (2) difficulties in controlling process parameters of the existing finishing techniques, and (3) inadequacy of most finishing techniques to improve geometric and form accuracies. Hence, we aim to tackle these issues with an emphasis on achieving precision and smart finishing via novel finishing tool development, finishing systems fabrication, process optimization, physical model establishment, integration of artificial intelligence and application exploration.
2. Data-driven manufacturing
Artificial intelligence provides a powerful tool to enable data-driven manufacturing. One example is the processing monitoring in complicated manufacturing process. We have been developing data-driven models for process monitoring, i.e., a broad echo state learning system (BESLS), which predicts the material removal rate (MRR) during polishing. This BESLS model has a flat structure and employs reservoirs with echo state properties to replace enhancement layer nodes of the conventional broad learning system. By fusing the time-domain and frequency-domain features of the polishing force as input signals, the proposed model yielded a 93.1% predicting accuracy of the MRR, outperforming the physics-based and phenomenon-based models. The proposed model also demonstrated remarkable computational efficiency with a testing time of 0.134 s, making it a potential candidate for online process control. The developed BLS algorithm also enables the online measurement of internal surface roughness, which is an important yet challenging task.
​
The impact of this research is that it can be applied for process monitoring and fault diagnosis in smart manufacturing. The data-driven measurement technique will be suitable for complex additive manufactured parts, for example, a thrust chamber the aircraft or rocket engine.
3. Functional surfaces manufacturing
We also work on manufacturing functional surfaces such as AM surfaces, freeform surfaces and microstructured surfaces which have wide applications.
The poor surface quality of AM parts greatly impedes their applications. Our method to address this issue is to study the mechanics during material removal of AM materials followed by introducing innovative methods for surface processing. Given the capabilities for complex geometry fabrication by AM, we study surface defects of sloped surfaces generated by SLM, which greatly affected the surface roughness with a strong dependency on slope angles. Additionally, we find that material removal rates during polishing closely follow the trends of surface roughness for which I developed an analytical microcutting model to successfully explain the dependency. Based on the developed understanding and experience in processing complex surfaces, we design a vibration-assisted conformal polishing tool to finish v-groove structures fabricated by SLM and also employed hybrid finishing methods to polish stent implants fabricated by SLM with an achievable nanoscale surface finish.
Robots are effective means for machine tools to adapt to the contours of freeform surfaces due to their high flexibility and system openness. While the concept may seem simple, it heavily involves tool path planning that also accommodates machine tool posture, orientation, and positioning for material removal regarding a given freeform surface. Achieving a tool path that conforms to the geometry of the freeform part alone is insufficient and we have developed solutions to achieve time-optimized tool path planning and sampling-based motion planning algorithms considering joint space and tool-tip kinematic constraints. These solutions greatly enhanced the machining efficiency and may be applied in robot-assisted additive manufacturing or polishing of surfaces with complex curvatures.
​
The generation of nanoscale roughness in polishing would not be possible without the fundamental understanding of ultraprecision machining (UPM) technology as an advanced manufacturing technology for generating superior surface finish and geometrical accuracy. In this field, we also have expertise in tool path planning of single-point cutting tools for the generation of spindle-shaped microstructures on rapidly solidified aluminum plates. These plates then went on to serve as precision molds for water harvesting based on PDMS casting with superior harvesting efficiencies three times that of an unmodified harvester. This research not only provides an attractive strategy towards strengthening water security with its cost-effective and zero-energy mechanism during water harvesting in high humidity conditions, but also demonstrates the wide applicability of UPM technology.
