Affiliations: Department of Mechatronics Engineering, National Changhua University of Education, Changhua, Taiwan, ROC
Correspondence:
[*]
Corresponding author. Yi-Cheng Huang, Department of Mechatronics Engineering, National Changhua University of Education, Changhua 500, Taiwan, ROC. Tel.: +886 4723 2105/Ext. 8138; E-mail: ychuang@cc.ncue.edu.tw.
Abstract: With the emergence of Industry 4.0, the development of smart machinery has become a goal of mainstream research. The computer numerical control (CNC) machine controller focuses on achieving excellent-quality finished products in a decreased amount of time, a stable surface roughness, and superior geometric accuracy. Therefore, a machining model based on the parameters of the CNC controller could be highly beneficial in industry. In this study, we analyzed the processing parameters of the CNC controller of Delta Electronics. A genetic algorithm (GA)-optimized general regression neural network (GRNN) prediction model based on Taguchi experimental data learning was constructed for a three-axis CNC machine. A fitness function with weighting value on developed GA-GRNN model was devised and navigated to deploy on different machining process needs. Each GA/GA-GRNN model finds a solution of five controller parameters inputs. Experiment results show the improvement of reducing machining time, jerk and corner error was achieved. The machining performance of each set of optimized parameters indicated that the parameter optimization system can assist users to obtain the CNC parameter combination that satisfies the processing requirements. This multi-objective GA/GA-GRNN model gives the intelligent CNC controller characteristics for recent smart manufacturing.
Keywords: CNC controller, machine tool, general regression neural network
