Surface roughness prediction of FFF-fabricated workpieces by artificial neural network and Box–Behnken method

Fused Filament Fabrication (FFF) or Deposition Modelling (FDM) three-dimension (3D) printing are rapid prototyping processes for workpieces. There many factors which have a significant effect on surface quality, including bed temperature, speed, and layer thickness. This empirical study was conducted to determine the relationship between above-mentioned average roughness (Ra). Workpieces of cylindrical shape were fabricated by an FFF system with Polylactic acid (PLA) filament. The measured at five different positions bottom top surface. A response (Box-Behnken) method utilised design experiment statistically predict response. total number treatments sixteen, while measurements (Ra 1 , Ra 2 3 4 5 ) carried out each treatment. settings factor as follows: temperature (80, 85, 90 °C), speed (40, 80 120 mm/s), thickness (0.10, 0.25 0.40 mm). prediction equation then derived from analysis. same set data also used inputs machine learning method, artificial neural network (ANN), construct roughness. Rectified linear unit (ReLU) activation function ANN. Two training algorithms (resilient backpropagation weight backtracking globally convergent resilient backpropagation) applied train multi-layer perceptrons. Moreover, neurons in hidden studied compared. Another interesting aspect this is that ANN based limited samples. Finally, errors compared, benchmark performance two methods: Box-Behnken