Cargando…
Model-Based Feedforward Control of Part Height in Directed Energy Deposition
Control of the geometric accuracy of a metal deposit is critical in the repair and fabrication of complex components through Directed Energy Deposition (DED). This paper developed and experimentally evaluated a model-based feedforward control of laser power with the objective of achieving the target...
Autores principales: | , , , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
MDPI
2021
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7826930/ https://www.ncbi.nlm.nih.gov/pubmed/33440854 http://dx.doi.org/10.3390/ma14020337 |
Sumario: | Control of the geometric accuracy of a metal deposit is critical in the repair and fabrication of complex components through Directed Energy Deposition (DED). This paper developed and experimentally evaluated a model-based feedforward control of laser power with the objective of achieving the targeted part height in DED. Specifically, based on the dynamic model of melt-pool geometry derived from our prior work, a nonlinear inverse-dynamics controller was derived in a hatch-by-hatch, layer-by-layer manner to modulate the laser power such that the melt-pool height was regulated during the simulated build process. Then, the laser power trajectory from the simulated closed-loop control under the nonlinear inverse-dynamics controller was implemented as a feedforward control in an Optomec Laser-Engineered Net Shape (LENS) MR-7 system. This paper considered the deposition of L-shaped structures of Ti-6AL-4V as a case study to illustrate the proposed model-based controller. Experimental validation showed that by applying the proposed model-based feed-forward control for laser power, the resulting build had 24–42% reduction in the average build height error with respect to the target build height compared to applying a constant laser power through the entire build or applying a hatch-dependent laser power strategy, for which the laser power values were obtained from experimental trial and error. |
---|