Dinh-Toan Vu

In the context of Industry 4.0, multi-sensor data plays a pivotal role in monitoring, analyzing, and optimizing product quality in real time. The ability to capture and process data from various sensors allows manufacturers to identify deviations, detect anomalies, and improve overall production efficiency. However, raw data collected during the injection molding process often contains redundant, irrelevant, or highly correlated features that can introduce noise and reduce the efficiency of predictive models. Without proper preprocessing, such data can lead to increased computational complexity and diminished model performance. To address these challenges, effective feature extraction techniques are essential for refining the dataset, minimizing prediction errors, and enhancing the interpretability of machine learning models. In this study, we compare two widely used feature extraction methods: Principal Component Analysis (PCA) and an Autoencoder (AE). The primary objective of this research is to assess the effectiveness of these feature extraction methods in monitoring the injection molding process and predicting product quality on an advanced machine learning model LSSVM. The experimental results presented in this study are useful in determining the suitability and disadvantages of each method, with the prediction accuracy of up to 99.62% for the extracted deep feature.

This study was performed to investigated the mechanism and stability prediction of flat-end milling process of SKD11 hard steel. The static milling force (MF) model, dynamic milling force (DML) model, milling vibration (MV) model, stability lobe diagram (SLD) model were developed and proposed in this study. The values of milling force coefficients (MLCs) and dynamic structure parameters (DSPs) were determined based on the combination of theoretical and experimental method. In static analysis, the relationship was very close to the linear regression with high determination coefficients (R2>97 %) in all three x, y, z directions. For experimental system of this study, the natural frequency of machine-tool system was almost half of natural frequency of machine-workpiece system. For the predicted results of SLD, the width of the stability region increased when the spindle speed increased. The machining processes were almost at the stable conditions with the axial depth of cut to be smaller than 0.3 mm at all values of spindle speed. The stable and unstable conditions in milling processes were verified with high precision for almost milling tests. With the points at the border between stable and unstable zones, the verified results had some differences between the predicted results using the proposed model and the analyzed results from experimental processes. These differences can be explained that there are many other parameters that have not been included in the predictive model such as heat, tool wear, friction, etc. In the applying of this study, it is the best way to choose the machining parameters at the absolute stable machining conditions.

This study presents an innovative approach to optimizing the milling process of C45 steel using a factorial design of experiments. The impact of key technological parameters, including cutting speed (Vc), feed per tooth (fz), depth of cut (ap), and lubrication conditions (dry and flood), on surface roughness (Ra) and Material Removal Rate (MRR) was thoroughly analyzed. Through the application of Analysis of Variance (ANOVA), significant factors and their interactions were identified, with fz and lubrication conditions showing the most substantial influence on Ra. The interaction between fz and lubrication condition was particularly notable, highlighting the importance of these parameters in achieving optimal surface quality. Multi-objective optimization was conducted using the desirability function method to balance the objectives of minimizing Ra and maximizing MRR. The optimal Vc and fz conditions under flood lubrication were found 200 m/min and 0.3 mm/tooth, respectively, achieving a desirability index of 0.801. Under dry lubrication, the optimal conditions were 200 m/min and 0.3 mm/tooth, respectively, with a desirability index of 0.803. These results demonstrate that both lubrication conditions can be effectively optimized to enhance machining performance. The findings provide a comprehensive framework for improving the milling process of C45 steel, contributing valuable insights into the effects of cutting parameters and lubrication conditions on surface quality and MRR.

This experimental study investigates the effect of the cutting parameters of the Electrical Discharge Machining (EDM) process of Ti-6Al-4V alloy material on surface roughness (Ra), cutting time (t) and Material Removal Rate (MRR) then solve the Multiple Objective Optimization Problem using separate Multi-Criteria Decision-Making (MCDM) methods namely Entropy-Weighted TOPSIS (E-TOPSIS), MOORA, SAW, VIKOR, and WPM. Focusing on nine 3-levels variants including Operating Voltage (OV), Pulse-On Time (Ton), Pulse-Off Time (Toff), Short-Circuit Off Time (AFF), Secondary Voltage (SV), Feed Rate (WF), Tension (WT), Water Pressure (WL) and Material Cutting Speed (F). Due to the large number of variants studied, the Taguchi L27 experimental design was chosen to reduce the number of experiments while still ensuring reliability in assessing the impact of technological parameters on responses in the study. The optimization results from the different methods indicated two distinct optimal outcomes. According to the E-TOPSIS, MOORA, and SAW, the optimal result is a Ra of 3.27 µm, a t of 7.37 min, and an MRR of 7.45 mm3/min. This result suggests a balanced and harmonious optimization among all criteria. On the other hand, the figures VIKOR and WPM methods are 2.87 µm, 9.57 min, and 5.74 mm3/min, respectively. These results indicate a higher priority for certain criteria, reflected in the lower Ra, longer cutting time, and a smaller MRR in comparison to the figures for the remaining MCDM methods. The different optimal results achieved by various methods highlight that each method is suited to and excels with different sets of values. Therefore, in each specific research or production process, comparing and choosing results calculated by different methods provides a comprehensive view, aiding in making appropriate decisions