Introduction: Predicting surface velocity accurately in Laser Powder Bed Fusion (LPBF) is essential for controlling melt pool dynamics. Things that minimize the defects and improve the quality of the product are really important. The usual fuzzy regression models often do not work well to address the issues with the product. Fuzzy regression models, like these, often fail to make the product better. We need to find ways to make the product have fewer defects and be of better quality. The product quality is what matters most. Fuzzy regression models should help improve the product quality. There is something about the numbers in LPBF data. It does not seem right. This study is about something called Asymmetric Triangular Intuitionistic, and it deals with LPBF data. LPBF data has a lot of uncertainty and things that do not match up which is called asymmetry. The people who did this study are trying to figure out LPBF data and its asymmetry and uncertainty. They are using Asymmetric Triangular Intuitionistic to do this. LPBF data is very hard to understand because of its asymmetry and uncertainty. Pythagorean Fuzzy Linear Regression models to improve prediction accuracy under uncertainty. Methods: The LPBF process parameters, such as laser power, scan speed and hatch spacing and layer thickness were changed into something called fuzzy for the LPBF process. This means the LPBF process parameters like laser power and scan speed and hatch spacing and layer thickness, are now used in a way for the LPBF process numbers using 98.5% and 95.5% confidence intervals. Two regression models-Intuitionistic Fuzzy Linear Regression (IFLR) and Pythagorean Fuzzy Linear Regression (PFLR)-were developed using fuzzy optimization and solved via MATLAB. We looked at how things were done by using things like MSE, R², adjusted R², and residual analysis to figure it out. We used these metrics like MSE and R² to see how good the performance was. The performance was assessed using metrics such as MSE, R², and adjusted R² and we did residual analysis too. Results: The IFLR model did well with an error rate of 0.0156, and it was able to explain a lot of what was going on with a score of 0.9234. The IFLR model also had a difference, between the highest and lowest values of 0.2345. On the other hand the PFLR model. The model produced an MSE of 0.0123 and an R² of 0.9456. This is really good because it had a spread of 0.1987. We saw that the model reduced the variance by 18.6% and 22.3%. The variance reductions are for the model. The model had these reductions. Discussion: Both models were really good at dealing with things that are not balanced and relationships that are not straightforward in LPBF data. Both models did a job with this. They could handle the relationships, in the LPBF data very well. The PFLR model did a good job of understanding things it had not seen before and making good predictions because the PFLR model is really good at showing lots of different things. The PFLR model is flexible so the PFLR model can represent things in various ways. Conclusion: The new asymmetric fuzzy models make a difference in how well we can predict things in LPBF processes. These asymmetric fuzzy models are really good at modeling things. That helps a lot. The asymmetric fuzzy models do a job of figuring out what is going to happen in LPBF processes. This is because the asymmetric fuzzy models can model things in a very accurate way. Uncertainty makes quality control tools really useful. They are also very helpful for optimization, in manufacturing. Advanced manufacturing uses these tools a lot because of the uncertainty.
Khan et al. (Thu,) studied this question.