論文

Selective laser sintering hybrid milling (SHM) is an advanced manufacturing technology that combines the flexibility of additive manufacturing with the precision of traditional machining. In SHM, instead of machining the products after they are completely formed using selective laser sintering (SLS), the milling process is conducted alternatively with the SLS. This allows direct fabrication of extremely complicated and high-precision products on the same workstation. However, shrinkage owing to SLS is a fundamental problem of this technology, because it creates significant dimensional errors in the final product. In this study, a method is proposed to calculate shrinkage error in SHM based on equations built from the data collected using F E A . Consequently, the calculation result is used to compensate the SHM shrinkage error. The experimental results of this study confirm that the error prediction model and the proposed compensation technique work efficiently in reducing SHM error.

Selective laser sintering hybrid milling (SHM) is an advanced manufacturing technology that combines the advantages of additive manufacturing and traditional machining. This technology is used to fabricate extremely complicated and high-precision products. The key factor of SHM is the milling process that is performed during the depositing process. However, the thermal shrinkage during the cooling process when depositing new layers on the ones that were machined is a fundamental problem because it creates deformation errors which significantly reduce product accuracy. In this study, the thermal deformation error in SHM was investigated using finite element analysis (FEA) and confirmed experimentally. The mechanism involved in the shrinkage in SHM was studied using three shrinkage rules. Moreover, a compensation technique was proposed with the aim of reducing the thermal deformation error. The results proved that compensation could significantly reduce the deformation error in SHM.

© 2019 Japan Society for Precision Engineering. All rights reserved. This study presents the development of an automatic, low-cost, fast, and reliable non-contact system for surface roughness estimation and chatter vibration identification based on machine vision. The developed system focuses on predicting two critical problems in the machining process, which are surface roughness and chatter vibration, using images as input. Deep learning with convolutional neural networks is integrated into the system to bypass the feature extraction method traditionally used in conventional vision-based roughness and chattel predictions. Two systems are proposed: Separate models that work specifically for each problem, and a combined model Thai aims to predict both problems in one process. Four deep learning architectures are proposed for both systems. The proposed models are built and tested on turned and milled surface datasets. A set of machining experiments are performed with various cutting conditions to generate the training and testing data. The result of the prediction model is then analyzed and compared with the measured data using a contact-based profilomctcr. The results indicate that the proposed system performs favorably in terms of accuracy and processing time, thus offering a promising alternative for quick inspection of surface quality.

In order to achieve high precision machining and high productivity, it is necessary to optimize machining data for a machine tool to be used, its control system, and shapes to be machined. Machining data are usually not optimized because it is generated by the CAM system as NC data based on general information. In the conventional process chain, even if the machining data is properly improved at the production site, there is no mechanism to reflect that in the previous process, so that the improvement information is not utilized. In this study, a shared database and a data model to share and inherit information in all processes from design to production is proposed in order to generate optimized machining data for the target machine tool and machining shape. The shared database and the data model enable non-expert operators to generate high-quality machining data like experts. In addition, it is also possible to reuse the information on the modified cutting conditions improved by experts at the machining site via a shared database. The effectiveness of the proposed shared database and data model is confirmed by experiments.

Laser metal deposition (LMD) is an additive manufacturing technique, whose performance can be influenced by a considerable number of factors and parameters. Typically, a powder is carried by an inert gas and sprayed by a nozzle, with a coaxial laser beam passing through the nozzle and overlapping the powder flow, thereby generating a molten material pool on a substrate. Monitoring the evolution of this process allows for a better comprehension and control of the process, thereby enhancing the deposition quality. As the metal additive manufacturing mechanism has not yet been elucidated, it is not clear how process parameters affect material properties, molding accuracy, and molding efficiency. When cladding is performed under uncertain conditions, a molded part with poor material properties and dimensional accuracy is created. In this paper, we propose a method for high efficiency molding by controlling the distance between the head nozzle and the molten pool in real time. The distance is identified by an originally developed sensor based on a triangulation method. According to the distance, the head nozzle is automatically controlled into the optimum position. As a result, an ideal molding process can be generated, so that high efficiency molding and high-quality material properties can be obtained. Experimental results show that continuing deposition at the optimum distance assists in achieving deposition efficiency and dimensional accuracy. According to the specific experimental results of this method, the modeling efficiency was increased by 27% compared to the method without correction, and the modeling was successful with an error within 1 mm.

Rapid estimation of machining time is necessary for flexible production scheduling and early responses regarding delivery date. It is also important for selecting the most suitable of a factory's many machine tools. Usually, machining time is estimated based on an NC program. However, this takes time to generate and its estimation accuracy is not ideal because it cannot consider the control characteristics of the machine tool. This study proposes a new method for rapidly estimating die and mold machining time without generating an NC program: inputting curvature and machining depth distributions into AI as color information.

Ablation processing using a femtosecond laser has the advantage of processing materials that are impossible to machine by conventional methods. However, it is difficult to predict the three-dimensional removal shape owing to a femtosecond laser irradiation and the current processing method, repeating through trial and error, is inefficient and expensive. For this reason, for processing materials to desired shapes using a femtosecond laser, development of a simulation system and a CAM (Computer-Aided Manufacturing) system are indispensable. In this paper, a simulation method for processing materials using a femtosecond laser is proposed and conducted to nano-polycrystalline diamond (NPD). It is confirmed that the proposed method can precisely simulate the femtosecond laser processing for three-dimensional shape generation. Therefore, the proposed method enables to predict the removal shape identified as the virtual tool shape for calculating a tool path on a CAM system. In addition, an automation process method including an on-machine real-time measurement system of the proposed method is described.

© 2018 Elsevier Ltd Especially with increased cutting forces using complex-shape tools such as long end milling tools, deflection in machining significantly affects the dimensions and geometry. This kind of error in end milling may be efficiently reduced by controlling the cutting parameters and properly selecting a suitable lubrication mode. The target of this research is to study the mechanism of deflection error and find the optimum condition for the cutting process to minimize deflection error. In this research, lubrication mode is considered one of four factors affecting deflection error in traditional flood lubrication, Minimal Quantity Lubrication (MQL), nanolubrication and dry cutting, which are used in the machining process with end milling tools. The three other parameters are axial depth of cut, radial depth of cut, and feed of cut. The workpiece material used in this research is AL 6061-T6, the cutting tool used is HSS end milling with diameter 4 mm, and the mode employed is side milling. The experiment is designed according to the Taguchi method, and it includes sixteen samples. The cutting fore and deflection error on a workpiece are recorded and then analyzed by S/N response analysis and analysis of variance (using Pareto ANOVA). The results indicate that radial depth of cut affects nearly 95% to deflection error. The effect of the other factors is not considerable. The optimum deflection error (31.22 μm) is attained at the lowest level of radial depth of cut (0.32 mm) and feed of cut (250 mm/min), and the second level of axial depth of cut (10 mm) and Minimal Quality Lubrication. A model of tool deflection named Instantaneous Deflection is defined and used to explain the phenomenon of deflection error. This model can explain the mechanism of deflection in end milling much better than average cutting force models. It proves that cutting force and distance between tool holder and cutting point are two parameters affecting deflection error. Between these two factors, cutting force has a much more significant effect. Cutting force is also proven to have a strong linear relationship with deflection error.

© 2020 Elsevier Ltd Existing computer vision methods to measure surface roughness rely on feature extraction to quantify the surface morphology and build prediction models. However, the feature extraction is a complicated process requiring advanced image filtering and segmentation steps, resulting in long prediction time and complex setup. This study proposes the use of convolutional neural network to evaluate the surface roughness directly from the digital image of surface textures. This method avoids feature extraction since this step is integrated inside the network during the convolution process. Five loss functions for the prediction models are selected and analyzed based on their suitability and accuracy. The predicted values obtained are compared to the actual surface roughness values measured using a stylus-based profilometer. The performance of the proposed model is evaluated for the prediction of the surface roughness of typical machining operations, such as outside diameter turning, slot milling, and side milling, at various cutting conditions.

© 2017 Scheduling in flexible manufacturing systems (FMS) is described as an NP-Hard problem. Its complexity has increased significantly in line with the development of FMS over the past years. This paper presents a non-dominated sorting biogeography-based optimization (NSBBO) for scheduling problem of FMS having multi loading-unloading and shortcuts infused in the reentrant characteristics. This model is formulated to identify the near optimal trade-off solutions capable of addressing the bi-objectives of minimization of makespan and total earliness. The goal is to simultaneously determine the best machine assignment and job sequencing to satisfy both objectives. We propose the development of NSBBO by substituting the standard linear function of emigration-immigration rate with three approaches based on sinusoidal, quadratic and trapezoidal models. A selection of test problems was examined to analyze the effectiveness, efficiency and diversity levels of the proposed approaches as compared to standard NSBBO and NSGA-II. The results have shown that the NSBBO-trapezoidal model performed favorably and is comparable to current existing models. We conclude that the developed NSBBO and its variants are suitable alternative methods to achieve the bi-objective satisfaction of reentrant FMS scheduling problem.