The challenge of maintaining consistent quality in injection molding is critical, yet conducting a comprehensive inspection is both costly and time consuming. Leveraging artificial intelligence, this ...study proposed using machine learning—specifically multilayer perception (MLP) models—to predict the quality of injection-molded parts. The accuracy of this approach largely relies on hyperparameter tuning, a process that can be cumbersome and suboptimal if performed through trial and error. The Taguchi method has the advantages of robustness, efficiency, and simplicity, and is a widely used robust optimization tool. However, this method assumes a linear relationship between factors, which limits the processing of complex systems where interactions between factors are nonlinear. Furthermore, the Taguchi method is sensitive to initial assumptions about factors and their levels, and the results may not reflect the true behavior of the system. To address this, a two-stage design-of-experiments method was devised that systematically identifies the optimal hyperparameter settings, including the maximum number of epochs, learning rate, momentum, activation function, minimum batch size, and numbers of hidden layers and nodes. The method is executed in two stages: (1) an
L
12
(2
1
× 3
5
) orthogonal array is used to identify the primary factors affecting model accuracy and (2) an
L
8
(2
3
) full-factorial experiment is conducted discover the combinations that yield the highest performance. Two experimental case studies, integrated circuit (IC) tray width prediction and optical component weight prediction, were used to validate the proposed method. The results revealed that the best hyperparameter settings resulted in validation and test accuracy of 96.83% and 95.30%, respectively, for IC tray width prediction. The average root-mean-square errors are 0.019 and 0.022 in model validation and test, respectively, for optical component weight prediction, with short computational time. The proposed method demonstrates how the systematic optimization of hyperparameters for MLP model can enhance the efficiency and stability of model training and can be used to advance quality control in the field of injection molding.
The aim of this work is the design of a pitot probe (PP) prototype in order to retard the cool down of the tip, in case of a heating element failure. The viability of operation in flight conditions ...is evaluated. The design consists of a redundant heating system incorporating phase change materials (PCM). Combining experimental observations of ice formation with the implementation of the conjugate heat transfer (CHT) model, with the addition of the heat release due to the phase change of the PCM, the numerical evaluation is developed. The modelling assumptions and numerical implementation of the phase change process are presented. Then, the selection an appropriate PCM is based on the low flammability and volume dilation and the quantitative effects of the material properties on the heat transfer. A commercial PCM solution based on salt hydrates was chosen as the most adequate for the design. The parametric design of the prototype, based on the design of experiment method and fractional factorial testing, is established. A multiple linear regression model was obtained in order to maximize the cooling retardation. The numerical simulations demonstrate that the prototype PP tip temperature remains 194 s longer above 0 °C than that of the conventional model analyzed.
•Design of Pitot Probe (PP) to retard cooling of the tip in case of a failure.•New: a redundant heating system incorporating phase change materials (PCM).•Selection of PCM and parametric design process is detailed.•Numerical model (already evaluated) is implemented to obtain a novel prototype PP.•Tip temperature remains 194 s longer above 0 °C than that of the conventional model.
The recirculation zone and the swirl flame behavior can be influenced by the burner exit shape, and few studies have been made into this structure. Large eddy simulation was carried out on 16 cases ...to distinguish critical geometry factors. The time series of the heat release rate were decomposed using seasonal-trend decomposition procedure to exclude the effect of short physical time. Dynamic mode decomposition (DMD) was performed to separate flame structures. The frequency characteristics extracted from the DMD modes were compared with those from the flame transfer functions. Results show that the flame cases can be categorized into three types, all of which are controlled by a specific geometric parameter. Except one type of flame, they show nonstationary behavior by the Kwiatkowski–Phillips–Schmidt–Shin test. The frequency bands corresponding to the coherent structures are identified. The flame transfer function indicates that the flame can respond to external excitation in the frequency range 100–300 Hz. The DMD modes capture the detailed flame structures. The higher frequency bands can be interpolated as the streamwise vortices and shedding vortices. The DMD modes, which correspond to the bands of flame transfer functions, can be estimated as streamwise vortices at the edges.
In the context of geotechnical and geological barriers, a thorough analysis of uncertainty and sensitivity is a crucial aspect of any physics-based performance assessment. While experimental data are ...scarce in actual waste repositories, large-scale experiments in underground research laboratories (URLs) provide such data that can be used to not only qualify THMC process models but also uncertainty assessment methodologies. In this paper, we adopt a Design of Experiments (DoE)-based history matching workflow – an approach popular in the oil and gas industry – and scrutinize its applicability for multiphysical analyses of nuclear waste disposal-related processes using synthetic experimental data. Based on an analytical solution of a coupled thermo-hydro-mechanical (THM) problem of a heat source embedded in a fluid-saturated porous medium mimicking a disposal cell in an argillaceous host formation, we discuss the adaptability of the workflow as a way to address parameter and model uncertainties for barrier integrity assessment. We thereby put particular focus on the relative importance of providing defined input parameter distributions for quantities generally afflicted with epistemic uncertainty and the constraints imposed by experimental (URL) or monitoring (repository) data. We found that once constraining data is available, the particular a priori distribution plays only a minor role for the outcome, such that we can conclude that the often unknown distributions can be substituted by uniform priors under such conditions. However, detailed knowledge of parameter distributions can increase the efficiency of the workflow significantly. We conclude that the presented workflow is particularly suitable for performing uncertainty quantification and sensitivity analysis for geotechnical applications where monitoring or other experimental data are available, as it allows us to deal with models of great complexity, epistemic uncertainty and it incorporates canonically to use of measured data in order to reduce uncertainty.
Micro-electro discharge machining (EDM) has a crucial role in fabrication of parts such as molds. One of the most challenging aspects of EDM process relates to flushing of debris. Increasing the ...material removal rate causes significant enhancement in EDM process. In this study, new numerical approach for enhancing flushing is proposed. A model of vibratory electrode and dielectric flow pattern is validated with theoretical and experimental results. Next, the effect of vibratory electrode geometry and its material properties on flushing of debris is investigated. To this end, design of experiment method is applied to find optimum vibratory electrode shape and material. At first step, vibrational analysis is performed for electrodes to find mode shapes and their corresponding frequency utilizing the finite element method. Vibrational results have good agreement with theoretical ones that are presented in literatures. Next, frequency response function analysis is done for vertical excitation force to find out displacements of electrodes in longitudinal modes. The results of these simulations are utilized to construct 3D computational fluid dynamic (CFD) models to find velocity distribution on gap space between electrode and work piece. Full factorial design is used to evaluate electrode’s parameters. The results from these simulations are studied to achieve best electrode shape that is able to flush the gap flow significantly. Finally, dielectric flow regime for optimum electrode shape is surveyed.
The generating heat from bearings and a motor of a built-in high rotating speed spindle of a machine tool has to be dissipated to attain high machining accuracy. In this study, a machine tool with a ...coolant cooling system for a high rotating speed spindle was investigated. The cooling channel designs in terms of channel pattern and dimensions by the design of experiment method (DOE) on the cooling effect of the spindle were studied. By the optimized cooling channel design, the cooling efficiency can be improved by 57.6% and the generating temperature in the spindle can be reduced by 25.9°C.
Laser hardening is one of the reasonable procedures utilized to improve surface mechanical and wear resistance of target materials. This process is easy to use for improving surface properties of ...complex shape components with minimum time requirement. The beauty of this process is that it is possible to improve only selective surface properties without altering the remaining surface properties of bulk material. Ck45 steel has various important properties like hardness, superior wear resistance and frictional performance etc. These properties are useful in various industrial applications. The existing study is an attempt to analyse results on surface properties with the aid of process parameters of a laser hardening process on hardness depth and microstructure of laser hardened Ck45 steel the use of a design of experimental approach. From this analysis, it can be summarized that the depth of the surface layer hardness is maximized by controlling the laser beam power, laser scan speed and standoff distance, which improves the service life of Ck45 steel components hardened by laser. Comparing the experimental consequences with the effects of L9 orthogonal array approach, the proportion error was observed to be negligible with a higher co-relation coefficient (r2) of 0.905.
Energy crises are one of the biggest crises of the world today. With ever increasing gadgets and machinery required for a comfortable lifestyle the demand for more and more energy is on its peak. So ...the present energy consumption scenario demands a much smarter and efficient energy storing technology. Henceforth, we need to find some alternate and most important the efficient energy storage methods. We all know that the conventional storage devices of electrical energy like lead acid, lithium ion, nickel cadmium etc. that are available today suffer many drawbacks. But the major problem with the high scale use of these storage devices is less efficient method of energy storage. Electrode materials along with all the other input parameters play the most important role in deciding the output parameters of the ultracapacitor. This paper presents the optimization of electrode parameters for development of stacked type of aqueous based ultracapacitor.
Purpose - The purpose of this paper is to investigate experimentally the effect of volume of casting, pouring temperature of different materials and shell mould wall thickness on the surface ...roughness of the castings obtained by using ZCast direct metal casting process.Design methodology approach - Taguchi's design of experiment approach was used for this investigation. An L9 orthogonal array (OA) of Taguchi design which involves nine experiments for three factors with three levels was used. Analysis of variance (ANOVA) was then performed on S N (signal-to-noise) ratios to determine the statistical significance and contribution of each factor on the surface roughness of the castings. The castings were obtained using the shell moulds fabricated with the ZCast process and the surface roughness of castings was measured by using the surface roughness tester.Findings - Taguchi's analysis results showed that pouring temperature of materials was the most significant factor in deciding the surface roughness of the castings and the shell mould wall thickness was the next most significant factor, whereas volume of casting was found insignificant. Confirmation test was also carried out using the optimal values of factor levels to confirm the effectiveness of this approach. The predicted optimal value of surface roughness of castings produced by ZCast process was 6.47 microns.Originality value - The paper presents experimentally investigated data regarding the influence of various control factors on the surface roughness of castings produced by using ZCast process. The data may help to enhance the application of ZCast process in traditional foundry practice.
The laser cladding process can be found in many different industrial applications. A lot of different material combinations were observed in recent years. For the application of laser cladded ...coatings in highly loaded areas, such as forming tool surfaces, the bonding characteristics between substrate and coating have to be evaluated and optimized. A special testing device is developed to measure the adhesive tensile strength of standardized laser cladded samples. To improve mechanical properties of the coating system within the process window, process parameters are tested and optimized by applying the design of experiment method. Results are presented from an iron based and a nickel based coating material on two different steel substrates.