Additive manufacturing technologies are widely used in engineering applications, enabling the rapid and low-cost production of complex geometries. Among these technologies, the fused deposition modelling (FDM) method is one of the most commonly used techniques, particularly in the production of polymers and polymer-based composite materials. However, the surface quality and dimensional accuracy of parts produced using the FDM method are significantly affected by production parameters.

In this study, the effects of different production parameters on the quality of samples were investigated using the FDM method with carbon fibre-reinforced polylactic acid (PLA-CF) composite polymer.

In the experimental studies, a total of 16 samples were produced using four different layer thicknesses (0.1, 0.15, 0.2 and 0.25 mm) and four different nozzle temperatures (220, 225, 230 and 235 °C). The surface roughness values of the samples, produced in dimensions of 30 × 50 × 3 mm, were determined by taking the average of ten measurements taken from different surface areas. Dimensional accuracy was evaluated by calculating the difference between the nominal dimensions of the samples and their actual measurements.

Analysis of the experimental results revealed that the lowest surface roughness value was obtained at a layer thickness of 0.1 mm and a nozzle temperature of 225 °C, with this value being 0.78 µm. It was observed that surface roughness values increased with increasing layer thickness. Furthermore, it was determined that the nozzle temperature improved surface quality up to 230 °C, but above this value, surface quality was negatively affected. The highest surface roughness value was measured at 6.67 µm with a layer thickness of 0.25 mm and a nozzle temperature of 235 °C.

The obtained surface roughness and dimensional accuracy data were also used as a dataset to evaluate the performance of machine learning algorithms. In this context, the prediction performances of different machine learning algorithms were compared to determine the most suitable algorithm.