• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.20-20
• /
• 2000

Back-propagation neural network (BPNN) is the most prevalently used paradigm in modeling semiconductor manufacturing processes, which as a neuron activation function typically employs a bipolar or unipolar sigmoid function in either hidden and output layers. In this study, applicability of another linear function as a neuron activation function is investigated. The linear function was operated in combination with other sigmoid functions. Comparison revealed that a particular combination, the bipolar sigmoid function in hidden layer and the linear function in output layer, is found to be the best combination that yields the highest prediction accuracy. For BPNN with this combination, predictive performance once again optimized by incrementally adjusting the gradients respective to each function. A total of 121 combinations of gradients were examined and out of them one optimal set was determined. Predictive performance of the corresponding model were compared to non-optimized, revealing that optimized models are more accurate over non-optimized counterparts by an improvement of more than 30%. This demonstrates that the proposed gradient-optimized teaming for BPNN with a linear function in output layer is an effective means to construct plasma models. The plasma modeled is a hemispherical inductively coupled plasma, which was characterized by a 24 full factorial design. To validate models, another eight experiments were conducted. process variables that were varied in the design include source polver, pressure, position of chuck holder and chroline flow rate. Plasma attributes measured using Langmuir probe are electron density, electron temperature, and plasma potential.

• Transactions of Materials Processing
• /
• v.25 no.3
• /
• pp.182-188
• /
• 2016

Flexibly-reconfigurable roll forming (FRRF) is a novel sheet metal forming technology conducive to producing multi-curvature surfaces by controlling the strain distribution along longitudinal direction. In FRRF, a sheet metal is shaped into the desired curvature by using reconfigurable rollers and gaps between the rollers. As FRRF technology and equipment are under development, a simulation model corresponding to the physical FRRF would aid in investigating how the shape of a sheet varies with input parameters. To facilitate the investigation, the current study exploits regression analysis to construct a predictive model for the longitudinal curvature of the sheet. Variables considered as input parameters are sheet compression ratio, radius of curvature in the transverse direction, and initial blank width. Samples were generated by a three-level, three-factor full factorial design, and both convex and saddle curvatures are represented by a quadratic regression model with two-factor interactions. The fitted quadratic equations were verified numerically with R-squared values and root mean square errors.

• Transactions of Materials Processing
• /
• v.26 no.6
• /
• pp.341-347
• /
• 2017

Flexibly-reconfigurable roll forming (FRRF) is a novel sheet metal forming technology conducive to produce multi-curvature surfaces by controlling strain distribution along longitudinal direction. Reconfigurable rollers could be arranged to implement a kind of punch die set. By utilizing these reconfigurable rollers, desired curved surface can be formed. In FRRF process, three-dimensional surface is formed from two-dimensional curve. Thus, it is difficult to predict the forming result. In this study, a regression analysis was suggested to construct a predictive model for a longitudinal curvature of FRRF process. To facilitate investigation, input parameters affecting the longitudinal curvature of FRRF were determined as maximum compression value, curvature radius in the transverse direction, and initial blank width. Three-factor three-level full factorial experimental design was utilized and 27 experiments using FRRF apparatus were performed to obtain sample data of the regression model. Regression analysis was carried out using experimental results as sample data. The model used for regression analysis was a quadratic nonlinear regression model. Determination factor and root mean square root error were calculated to confirm the conformity of this model. Through goodness of fit test, this regression predictive model was verified.

• Journal of the Korean Society for Precision Engineering
• /
• v.32 no.2
• /
• pp.159-166
• /
• 2015

Slot-die coating is well known technique to guarantee a uniformly coated layer and is compatible with roll-to-roll process. In actual roll-to-roll slot-die coating process, the lateral difference of coated layer thickness is observed. An experimental study was performed to improve the coating quality. Coating speed and coating gap were selected as the experimental factors. A full factorial, statistical method was conducted to optimize the process conditions. Based on the results of repeated experiment, the lowest deviation of lateral thickness (700 nm, <10%) was achieved at 10 m/min coating speed and $300{\mu}m$ coating gap. This result has significance because such optimized process guideline can be utilized with all process improvement in various coating applications.

• Food Science and Biotechnology
• /
• v.14 no.6
• /
• pp.747-751
• /
• 2005

Microencapsulation of pine flavors was investigated to determine the optimum wall material and spray drying condition. ${\beta}$-Cyclodextrin, maltodextrin, and a 3:1 mixture of maltodextrin and gum arabic were evaluated as wall materials. The latter mixture was determined to be the best wall material based on dispersion capacity and flavor yield. Spray drying effectiveness was evaluated using a $3^3$ fraction factorial design and statistical analysis. The optimum operation condition was an inlet air temperature of $175^{\circ}C$, inlet airflow rate of $0.65\;m^3/min$ and atomizing pressure of 180 kPa, which resulted in a 93% flavor yield. The best particle shape observed by SEM was a round globular shape obtained under the above spray drying condition, whereas lower temperatures and higher inlet airflow rates resulted in initial and full collapses, respectively. The round globular shapes remained stable for at least one month.

• Korean Journal of Materials Research
• /
• v.25 no.7
• /
• pp.322-329
• /
• 2015

The magnitude of wear should be at a minimum for numerous automobile and aeronautical components. In the current work, composites were prepared by varying the cenosphere content using the conventional stir casting method. A uniform distribution of particles was ensured with the help of scanning electron microscopy (SEM). Three major parameters were chosen from various factors that affect the wear. A wear test was conducted with a pin-on-disc apparatus; the controlling parameters were volume percentages of reinforcement of 5, 10, 15, and 20%, applied loads of 9.8, 29.42, and 49.03 N, and sliding speeds of 1.26, 2.51, and 3.77 m/s. The design of the experiments (DOE) was performed by varying the different influencing parameters using the full factorial method. An analysis of variance (ANOVA) was used to analyze the effects of the parameters on the wear rate. Using regression analysis, a response curve was obtained based on the experimental results. The parameters in the resulting curve were optimized using the Genetic Algorithm (GA). The GA results were compared with those of an alternate efficient algorithm called Neural Networks (NNs).

• Journal of Welding and Joining
• /
• v.35 no.2
• /
• pp.35-42
• /
• 2017

The GMA welding process involves large number of interdependent variables which may affect product quality, productivity and cost effectiveness. The relationships between process parameters for a vertical weldment and mechanical properties are complex because a number of process parameters are involved. To make the vertical-position welding, a method that predicts bead geometry and accomplishes the desired mechanical properties of the weldment should be developed. In addition, a reliable welding process and conditions must be implemented to reduce weld structure failure. In this study, the welding process analysis of investigates the interaction between the heat input and welding parameter(Welding current, Arc voltage, Welding speed) for predicting the weldment hardness.

• Journal of the Korean Ceramic Society
• /
• v.43 no.9 s.292
• /
• pp.519-526
• /
• 2006

A full-factorial design of experiments with three input factors and two levels for each factor including center points was utilized for the preparation and characterization of twelve types of $BaTiO_3$ slips for tape casting. Ceramic powders with different particle sizes, different milling methods such as high energy milling and conventional ball milling, and two types of dispersant with different polymeric species were chosen as input factors in order to investigate their effects on slip and on green tape properties. Tape casting, a small rectangular-shaped K-square preparation, characterization and quantitative data analysis using statistical software were followed. Ceramic powder was the most significant among three input factors for the output responses of slip viscosity and green tape density, showing more favorable results with large particles than with very fine ones. In addition, high energy milling for only 30 min was more efficient than 24h of conventional ball milling in terms of powder dispersion and milling. The optimum condition based on the experimental results was a slip exposed to high energy milling with large ceramic particles along with a methylethyl acetate dispersant.

• Journal of the Korean Society for Precision Engineering
• /
• v.33 no.8
• /
• pp.611-616
• /
• 2016

Computational fluid dynamic simulation based on the ABAQUS software was conducted to observe the inside flow of slot-die nozzle. The slot-die nozzle was modeled as 3-dimensional structure and three significant parameters were determined: inlet velocity of fluid, reservoir angles, number of strips none of which have been mentioned previously in the literature. The design of experiment, full factorial analysis was performed within determined design and process levels. The simulation result shows the inlet fluid velocity is most significant factor for the flows of inside nozzle. As an interaction effect, reservoir angle is closely related with number of strip that should address when the nozzle is designed. Moreover, the optimized values of each determined parameter were obtained as 35 mm/s of inlet velocity, 3 of strip numbers, and $22^{\circ}$ of reservoir angles. Based on these parameters, the outlet velocity was obtained as 0.53% of outlet uniformity which is improved from 8.67% of nominal results.

• Journal of the military operations research society of Korea
• /
• v.33 no.1
• /
• pp.59-74
• /
• 2007

Computer simulation models are in general quite complex and time-consuming to run, and therefore, a simpler meta-model is usually constructed for further analysis. In this paper, JANUS, a war-game simulator, is used to describe a certain tank combat situation. Then, second-order response surface and artificial neural network meta-models are developed using the data from eight different experimental designs. Relative performances of the developed meta-models are compared in terms of the mean squared error of prediction. Computational results indicate that, for the given problem, the second-order response surface meta-model generally performs better than the neural network, and the orthogonal array-based Latin hypercube design(LHD) or LHD using maximin distance criterion may be recommended.