The main aim of the investigation was to predict chip form based on machining parameters and surface roughness. Straight turning of mild steel and AISI 304 stainless steel were performed. Spindle speed, feed rate, depth of cut and surface roughness of the material were used as inputs. Computational intelligence techniques could be used for the prediction process. In this article support vector regression (SVR) was applied for the chip form prediction. The SVR model was compared with other computational intelligence models like artificial neural network (ANN) and genetic programing (GP) techniques as benchmark models. The crucial aim of the study was to predict favorable and unfavorable chip form according to the machining parameters. By the way one should make optimal machining conditions in order to avoid unfavorable chip form. Based on the results, SVR (R 2 : 0.9682) model outperformed ANN (R 2 : 0.8367) and GP (R 2 : 0.7753) model for the chip form prediction..

No abstract

Adaptive neuro fuzzy network or ANFIS could be used for different aspects of prediction during productive development. In this study ANFIS network was appliced for prediction of bending and thickness of shaped surface by laser formation. Laser manufacturing represent imporant part in new productive development. Shaped surface modeling by laser forming process needs many irradiations along surface paths with different parameters of heating. This surface requires thickening and bending in order to get required shapes. In this study was attempted to analyze the process parameters influence on the bending and thickening of the shaped surface by laser. The used inputs were circular and radial laser scan, laser spot diameter, laser power and scan speed. The selection procedure can produce results to simplify the shaped surface forming. Finally it was attamted to determine how services as added values have infleunce on the enterpreneusthip activity for laser manufacturing business.

The main aim of the study was to analyze power coefficient of tidal turbine. The analzying was performed based on input-output data pairs. To arrange the data pairs measurements were performed on the tidal turebine. Adaptive soft computing methdology was used for estimation of relationship between the data pairs. The soft computing methodoogz was afterwards used for prediction of the power coefficient of tidal turbine based on the learined knowledge about the data pairs. During measurement procedure three inputs and one output were considered. The inputs are tip speed ratio-TSR, swap area of the turbine and time step. The output is power coefficient. The soft computing approach, namely, adaptive neuro fuzzz inference system-ANFIS was used for prediction of the power coefficient. Finally obtained results were compared with classical neural networks.

In this study was analyzed the influence of laser welding parameters on output parameters prediction. Adaptive neuro-fuzzy inference system (ANFIS) was applied for the variable selection process to determine the parameters influence on the lap-shear strength and weld-seam width prediction. The used inputs were: laser power, welding speed, stand-off distance and clamping pressure. Experimental test were used to acquire the training data for the ANFIS network. The ANFIS network was used to predict the lap-shear strength and weld-seam width according to the input variables separately. Root mean square error (RMSE) was used as statistical indicators for comparison. The results from this study could be used as benchmark results in order to improve the laser welding process.