Anomaly detection for cutting processes is an important topic in manufacturing. Detecting cutting tool anomalies on real time enable us to prevent products from being discarded. This research proposes an anomaly detection method for cutting tool monitoring by dynamical fluctuation analysis on topological phase space.

As labor shortages and technology transfer issues become more serious, demand for advanced processing automation technology is increasing. Cutting force in particular is said to be an effective indicator for machining monitoring, but there are challenges in measuring it, such as high cost and difficult installation. In this study, to solve these issues, we propose an AI technology that utilizes sensorless cutting force estimation and servo information during cutting. Through this, we aim to realize a low-cost, highly accurate cutting force estimation technology that does not rely on conventional measurement methods.

Understanding measurement uncertainty is the key to obtain reliable results. Uncertainty from a well-defined physical systems can be systematically propagated according to GUM. However, when measurement systems integrated with deep learning, it introduces new challenges due to mathematical & statistical complexity. This paper explores the principle to propagate the measurement uncertainty through deep learning.

The goal is to accurately identify interference fringes buried in noise. The training target was an image. The images were divided into two categories and three deep learning models were used: Google's Teachable Machine, ResNet50, and EfficientNet. We compare the accuracy and time efficiency of these models and provide insight into the trade-offs between general-purpose and task-specific neural network models.

Image reconstruction via scattering medium has evolved gradually for several decades. Through the advancement in technology particularly in physics, optics, and computational imaging techniques, it has brought significance imaging advancement in the fields of biomedical sciences, optical imaging, optical communications, and mesoscopic physics. Existing researches tackling scattering medium problem utilizing deep learning in various application has been deployed. However, those researches are constrained only to 2D (+phase). Here, we proposed a 3D speckle decoder to reconstruct the phase information retrieved from the speckle patterns in x, y, and z direction. To validate the ability of the 3D speckle decoder we test it both on 2D and 3D images utilizing MNIST and MNIST with SLM displacement in z-direction data sets respectively.

We develop a particle position detector which can work highly stably for a long term from several months to several years in a severe environment of high magnetic field and a vacuum. It has a cylindrical shape with a diameter of 700 mm and a height of 1 m. In order to ensure the detector not to be deformed, we develop an alignment monitor for being installed into the detector. The alignment monitor consists of length measurement paths of multiple absolute distance interferometers by refereeing an optical frequency comb. Here, we evaluate uncertainty factors for the interferometers and consider methods for their elimination.

The management of recesses in Cu-CMP processes needs to be performed at the nanometer scale. Currently, contact-based measurements such as AFM are employed, but the lengthy measurement time poses a significant challenge. Therefore, we report on an investigation into the sensitivity of optical measurement by examining the relationship between variations in shape parameters such as height and optical parameters using the RCWA, a vector model analysis method.

Unlike conventional autocollimators that use collimated light, we propose a new sensor that can measure angular displacement in addition to optical axis position by applying the principle of second harmonic generation in nonlinear optical crystals to confocal probes as a new angle measurement method using converged light that can measure angle in a local area. A prototype optical system was constructed. We construct a prototype optical system and experimentally investigate the proposed principle.

Nanoscale phenomena such as attached and/or detached nano-objects and/or surface damage caused by fluid pressure waves (density changes) generated near the polished surface. However, pressure waves near the surface are not measured and the mechanism is not clear. In this article, we experimentally considered a method to measure the refractive index near the surface using decreased of optically internal reflection intensity due to the tiny surface roughness, which would change with fluid density.

In wet processes such as polishing or cleaning, extremely small interaction forces act between nano-abrasive particles and the substrate surface. Understanding these interaction forces enables efficient determination of conditions for high-precision processing. In this article, we enhanced magnetic force using multiple twisted coils for measuring interaction forces at closer proximities to the surface and compared the results with those obtained using a single coil under identical applied current conditions. It was suggested that the enhanced attraction allowed the particles to approach the substrate surface more closely, enabling the measurement of interaction forces in cases of ultra-close nano-abrasive particles.

In our previous works, a method of electrochemical machining (ECM) using bipolar pulses and an auxiliary electrode has been proposed to improve machining accuracy, through reducing stray corrosion by utilizing intensively generated hydrogen bubbles. Additionally, for long-term machining applications (such as drilling and milling), it is crucial to apply external electrolyte flow to refresh the electrolyte and remove sludge products. However, external electrolyte flow affects the distribution of intensive hydrogen bubbles, thereby impacting the machining results. In this study, the effect of external electrolyte flow on machining characteristics is experimentally investigated.

Although micro-holes are widely used in aerospace and other industry fields, drilling micro-holes with large depth-to-diameter ratio is still one of the challenges in manufacturing. Electrical discharge and electrochemical hybrid machining (hybrid EDM-ECM), as a nontraditional machining process used for machining difficult-to-machine materials, has performed well in the field of micro-hole drilling. In this paper, a low conductivity electrolyte, which is easy to decompose and produce gases, is used as the working fluid to investigate the impacts of process parameters such as tool feed rate, capacitance value and open circuit voltage on micro-hole drilling performance.

When DC voltage is applied to silver ion doped glass at high temperatures below the softening point, silver precipitates are formed in the glass. In this study, to discuss the temperature dependence of the growth rate of the precipitates, we dynamically observed the silver precipitation behavior by applying a voltage to glass heated by a plate heater. The growth rate of silver precipitates decreased with increasing temperature. Therefore, this is considered that the crack formation in the glass and the ion diffusion coefficient affects the precipitation behavior.

To achieve high-precision and fine machining, dielectric oil have recently begun to be used in wire EDM. However, the effects of the properties of dielectric oil on wire EDM characteristics have not yet been sufficiently clarified. In previous studies, the impact of volume resistivity and viscosity of dielectric oil on wire EDM characteristics has been examined. In this report, the fundamental study was conducted on dielectric oil cooling effect on wire EDM characteristics.

In wire electrical discharge machining (EDM), bubbles and debris generated during machining influence the machining characteristics. However, their effects and exclusion behavior have not yet been clarified sufficiently. In this study, the flow of working fluid in a machined kerf in the presence of bubbles was discussed by Euler gas-liquid two-phase flow analysis. Furthermore, the debris exclusion behavior was discussed in detail by Lagrange method with working fluid flow field obtained by Euler two-phase flow analysis.

In this study, we investigated the machining characteristics of planar and circular grooved workpieces using a newly developed polishing wheel which can simultaneously apply magnetic field and electric field as a planar polishing method using Magnetic Compound Fluid. The tool has two ring-shaped permanent magnets arranged in the same pole orientation, and an electric field can be applied between the magnets. The effects of applying magnetic field alone or magnetic field and electric field simultaneously on machining characteristics such as material removal amount, surface properties, and torque acting with the tool were clarified.

In recent years, demand for micro-complex shaped parts has been increasing in various fields. In addition, there is a demand for precision machining technology to finish micro-complex shaped parts at the nano-level and with high efficiency. However, conventional polishing techniques have problems in that it is difficult to polish complex shapes precisely because the polishing does not reach the complex shapes. In this study, we attempted to polish the inside of a groove on a flat aluminum plate using a magnetic polishing method that utilizes a fluctuating ground field. The experimental results show that precise polishing of the inside of the grooves can be achieved.

This study presents how to create contextual and interoperable data for smart manufacturing. While the manufacturing industry generates vast amounts of data, it is often fragmented and requires additional context to be effectively utilized in a smart manufacturing environment. Current data generation tools, while powerful, are often overly rigid and require high levels of expertise, making them unsuitable for real-world applications. This study proposes a framework that leverages natural language annotations to consolidate fragmented data generated from various entities, such as manufacturing processes, materials, and physical phenomena, as linked data. This innovative approach adds context to datasets, improves usability, and provides datasets in a form that is understandable to both humans and machines. By emphasizing ease of use and flexibility, the framework aligns with smart manufacturing principles and establishes the foundation for an extensible system that adapts to the evolving needs of smart manufacturing. This user-friendly nature of the framework makes the audience feel comfortable with its implementation, thereby promoting data sharing and utilization.