|
Investigation and machine learning prediction of one-way progressive pump abrasive flow machining prototype: SKD61 (AISI H13) hot work steel |
|---|---|
| รหัสดีโอไอ | |
| Creator | Parinya Srisattayakul |
| Title | Investigation and machine learning prediction of one-way progressive pump abrasive flow machining prototype: SKD61 (AISI H13) hot work steel |
| Contributor | Theerapong Maneepen |
| Publisher | Khon Kaen University, Thailand |
| Publication Year | 2568 |
| Journal Title | Asia-Pacific Journal of Science and Technology |
| Journal Vol. | 30 |
| Journal No. | 4 |
| Page no. | 3 (13 pages) |
| Keyword | Abrasive Flow Machining, Surface Roughness, Design of Experiment, Machine Learning |
| URL Website | https://apst.kku.ac.th/ |
| Website title | https://apst.kku.ac.th/investigation-and-machine-learning-prediction-of-one-way-progressive-pump-abrasive-flow-machining-prototype-skd61-aisi-h13-hot-work-steel/ |
| ISSN | 2539-6293 |
| Abstract | The objectives are to study the effects of tooling pressure and machining time on surface roughness regulatory affairs (Ra) of hot work steel SKD 61 involving an abrasive flow machining (AFM) polishing prototype and to compare the design of experiments (DOE), factorial regression, and predictive machine learning (ML). The research steps include defining the objectives, identifying the important factors and their levels, designing the experiments, conducting experiments and collecting data, and analyzing the results statistically. The ML process was performed with a program, the experimental results were compared and concluded. A progressive pump (NETZSCH) was used to conduct the experiments on the AFM curved samples. Experiments were conducted to investigate machined, hardened, polished samples with abrasive papers from #180 to #1200 and were measured with the initial Ra values against the final value. The process parameters were pressure 1, 2, and 3 bar; sample hardness SKD61; 45±2 HRC; abrasive particle size (Al2O3) 5.0 microns (μm) (concentration 50% by weight, silicone oil). The results showed that 3 bar and 20 min, pressure, and cycle time consequently gave the best results. The average Ra of the workpieces was 0.057 to 0.042 microns, and the delta between surface roughness (SR) was 0.006 to 0.012 μm. The comparison of experimental and prediction results using RapidMiner ML showed only minor differences, indicating better precision control for industrial applications. |
