A machining center is a common machine tool that can machine complex free-form surfaces such as the press mold, automobile’s cam profile. When these curved surfaces are machined by the conventional machining center and/or the grinding machine, it takes lot of time to complete with enough accuracy. In order to solve this problem, this study aims not only to improve the machining accuracy of this curved surface but also to make it possible to shorten the machining time using a CNC lathe. In this study, the NACS-Turning (Non-axisymmetric curved surface turning) method that we proposed originally was used. The spindle of C-Axis, the moving table with the cutting tool of X₁-Axis, the other counter moving table of X₂-Axis (without involved in shape creation, only used to suppress inertial force), the head stock of Z-Axis, and the tool rotation axis of B-Axis are adapted. In this 4-Axis machining method, all the axes are controlled with the synchronized manner. This method can machine the same rotational position of workpiece with the same cutting edge of the rotary tool. As an experiment, we compared 4-Axis machining with synchronized spindle and tool rotation, 3-Axis machining, and machining without synchronous rotation. Our new 4-Axis machining method with synchronized spindle and tool rotation reduced the form error from 323 ㎛ to 247 ㎛ and the surface roughness from Rz 9.6μm to Rz 3.7μm. Moreover, the repetitive machining resulted in error of 99.6μm and a surface roughness of Rz2.9μm.

Recently, the development of high-hardness coating materials has made it possible to cut high-hardness steels, but machining efficiency is limited. Production efficiency can be improved if high efficiency machining can be achieved while maintaining the surface accuracy required for finish machining of hardened steel. This requires optimization of tool material and cutting environment. In this study, driven rotary machining of hardened steel with a (Ti,Al)N coating with high Al content was conducted to investigate the effect of different cutting environments on tool life. The results showed that under micro dry ice powder system (DIPS) at a cutting speed of 3.34 m/s, tool life was extended by using (Ti,Al)N coated tools with the highest Al content compared to the Dry environment. However, thermal crack occurred due to the cooling effect of DIPS at high speed of 5.00 m/s, and tool life is significantly shortened.

Workpiece rigidity during end-milling processes decreases monotonously. Reduction rates of the rigidity become large when the end-milling operations have large cutting stock such as carve out machining. It becomes popular to utilize side-face clamping methods because of large accessibility of cutting tools. However, workpiece rigidities directly affect to clamping force and workpiece deformation when workpieces are clamped with side-face clamping method. This research deals with transitional workpiece deformations considering workpiece rigidity reduction. At first, measured results of clamping force reduction during machining operations are investigated. Secondly, measured workpiece profiles during machining operations are investigated. Finally, FEM analysis of workpiece deformation considering workpiece change is discussed. From the results of the evaluation, it becomes clear that it is necessary to consider the influence of transitional workpiece rigidity to achieve reliable estimation of machining accuracy.

A capacitive MEMS gas sensor is fabricated by transfer printing of Au thin film and inkjet printing of molybdenum disulfide (MoS₂) nanoparticles. The transferred Au thin film is formed to be a comb electrode with a width of 100 μm and a thickness of 50 nm, and the comb-like Au electrode surface becomes hydrophilic by UV irradiation. The MoS₂ particles are deposited on the comb-like Au electrode as a sensing element, where the applied voltage and the pulse width of the piezoelectric inkjet head are adjusted to make a drop velocity of 6 m/s. The UV irradiation helps the MoS₂ particles to be deposited uniformly while suppressing undesired aggregation of the particles. The fabricated sensor composed of the comb-like Au electrode and the film of MoS₂ particles is placed in a chamber for testing, and the capacitance is measured with exposure to ethanol gas. It is demonstrated that the capacitance increases with exposure to ethanol gas and subsequently decreases after air-exposure. It is reasonable that the capacitance change is due to the adhesion of ethanol gas molecules to the MoS₂ particle film, and that the fabricated structure works as a capacitive MEMS gas sensor. It is also found that there are variations in response due to the surface morphologies of the MoS₂ particle film.

Virtual sensor, which is one of the applications of the digital twin, is effective for estimating physical quantities that are difficult to measure. This paper deals with estimation of real-time tool edge temperature by virtual sensor. The tool edge temperature is estimated by inputting the temperature data of the endmill shank obtained from the pyrometer installed in the machine tool to the simulation model created in cyber space. The simulation performed in cyberspace is an unsteady heat transfer analysis, and a state space model with a reduced order of the finite element model using Krylov subspace method is adopted as the simulation model in order to allow real-time simulation. In addition, unknown heat input to the cutting edge was calculated by state feedback control using the temperature data obtained from the pyrometer and the simulation model. The tool edge temperature during cutting was estimated in real time using the developed virtual sensor with the heat transfer analysis block and the input estimation block. The estimated tool edge temperature changed significantly in response to changes such as chipping of the tool edge. The estimated temperature and the measured temperature by the thermocouple method generally matched.

The purpose of this paper is to examine which coating material on tool edges is best for cutting spheroidal graphite cast iron containing 35% Nickel (Ni) and to understand the relation between friction properties of the coatings and cutting ability. Heat-resistant alloy materials, such as Ni-based alloy steels and Ni-based alloy cast irons containing a large amount of Ni, are known as difficult-to-cut materials due to high high-temperature strength, low thermal conductivity, and work hardening. In this study, austenitic spheroidal graphite cast iron containing 35% Nickel (FCDA-Ni35) was used as a work material, and shaper machining tests were performed for evaluating cutting abilities of TiN and CrN-coated tools. Furthermore, the friction properties of each coating were compared and evaluated by a ball-on-plate test. It was found that the ploughing force component of the CrN-coat was smaller, and the chip thickness was thinner than that of the TiN-coat, so the cutting ability was good. As a result, the CrN-coated tool has smaller cutting resistance and ploughing force component than the TiN-coated tool, and the friction coefficient of the rake face is smaller. Furthermore, the CrN-coated tool has stable chip formation, thin chip thickness, and good cutting ability on cutting for FCDA-Ni35.

Grinding is mostly utilized as finishing process in production. Thus, it is required to maintain performance of grinding wheel for preventing grinding failures such as grinding burn, loading, chatter marks, and so on. Although regular dress-interval for grinding wheel is adopted in practice, if degradation of grinding performance can be detected in process, not only the best dress interval can be determined, but also no-failure grinding can be realized.

Therefore, in this study, super abrasive wheel with new disk design is made smart by means of attaching sensors on disk of wheel to monitor load of grinding wheel. Also, in order to realize in-process grinding control, tool-work-velocity-control-system which is consist of microcomputer and original electro board has been developed. This system can control grain depth of cut by means of controlling relative velocity between grinding tool and workpiece. Utilizing the smart super abrasive wheel and the tool-work-velocity-control-system, cylindrical grinding experiment is executed.

As the result, the combination of those developed items shows availability that they can provide continue of grinding process without any failure even if sharpness of grinding tool is degraded.

This study unveils the effect of ion beam (IB) irradiation on mechanical properties and adhesion of titanium nitraide (TiN) thin films on ultrafine-grained WC-Co cemented carbide substrates. TiN thin films are used for improving wear resistance and lubricity of molds and cutting tools consisted of WC-Co cemented carbides. It is known that the ion beam assisted deposition (IBAD) improve the adhesion strength between the coatings and substrates. In this paper, TiN coatings were deposited on the ultrafine-grained WC-Co cemented carbide substrates by IBAD method that titanium was evaporated by an electron beam (EB) and nitrogen gas was introduced with/without IB. The chemical composition of TiN coatings deposited by IB were investigated by EPMA, XRD and XPS. The nitrogen ratio of TiN increased by IB irradiation in deposition process. The mechanical properties of TiN coatings with/without IB were evaluated by nanoindentation tests and ball-on-disc tests. Dynamic hardness and Young's modulus of TiN deposited by IBAD was 100% and 50% higher than that by conventional vapor deposition (VD), respectively. Friction coefficient of TiN deposited by IBAD was lower than that by VD. The adhesion of TiN coatings was measured by scratch tests. The adhesion strength of TiN deposited by IBAD was 21% higher than that by VD.

In order to design fast reactors, it is necessary to consider high cycle fatigue of structural materials up to 1×10⁹ cycles. Since the fatigue usage factor (UF) is calculated down to the order of 0.01 in fast reactor design, it is necessary to develop fatigue curves applicable up to 1×10¹¹ cycles to evaluate high cycle fatigue at 1×10⁹ cycles. In this study, the applicability of the best-fit fatigue curve of JSME up to 1×10¹¹ cycles was verified in order to develop a high cycle fatigue evaluation method for Mod.9Cr-1Mo steel, which is a candidate material for fast reactor structural materials. In the applicability verification, high cycle fatigue tests were first conducted at 500 °C, 550 °C, and 600 °C under strain-controlled conditions, and ultrasonic fatigue tests were conducted at 550 °C. The test results were then evaluated and showed that there was no rate dependence and no difference by product types for high cycle fatigue strength of Mod.9Cr-1Mo steel at high temperatures. It was also clarified that there was no difference in high cycle fatigue strength of Mod.9Cr-1Mo steel at high temperatures even if the failure mode changed from surface origin to internal origin. A best-fit fatigue curve was developed from test data obtained in high cycle fatigue tests up to 1×10⁹ cycles. And based on the high cycle fatigue mechanism discussed from the ultrasonic fatigue test results, the best-fit fatigue curve of JSME was confirmed to be applicable to Mod.9Cr-1Mo steel up to 1×10¹¹ cycles regardless of product types.

The co-sintering of multilayer ceramics, produced by tape casting, is a cost-effective process for manufacturing electrodes in solid oxide fuel cells (SOFCs). Hence, understanding the deformation kinetics of multilayer ceramics during co-sintering, originating from the different shrinkage rates of individual layers, has become critical. In-situ monitoring of the sintering process using an optical device is a promising non-contact method that enables the deformation tracking of both monolayer and multilayer ceramics. However, the material emits thermal radiation at various levels during sintering, which leads to a time-consuming image processing procedure, because the intensity of thermal radiation changes significantly when shifting from low to high temperatures. This study presents deep learning techniques that employ convolutional neural networks to segment and detect objects in the image processing associated with in-situ monitoring. We verify that the DeepLab V3+ network accurately segments the sintered body, even when images are severely affected by mixed brightness or noise. Furthermore, the presented network can also quantify shrinkage profiles over a wide temperature range within a short time. In addition, we demonstrate that the YOLO V4 network can monitor the warpage behavior of co-sintered laminates by evaluating the curvature profile over a broad range of temperatures without requiring image binarization, despite the limitations of curvature detection. Finally, we demonstrate how the developed networks can potentially be utilized to predict the time evolution of warpage deformation in SOFC anode/electrolyte bilayer systems.

The finite-element simulation using three yield criteria (Yld2000-2d, CPB06 with and without considering the SD effect) applying isotropic or anisotropic hardening rule of a commercially pure titanium sheet drawing were carried out, and their accuracies were evaluated by comparing them with experimental result. In experiments, the maximum cup height appears at 43 angle to the rolling direction, and there are positions with locally minimum cup heights (valley) around 0° and 90°. The shape of valley around 0° is almost flat which means the heights from 0° to 14° are almost same, unlike around 90° with round shape. The accuracy of simulation was greatly improved by applying anisotropic hardening rule in each yield criteria. Moreover, that using CPB06 with considering the SD effect applying anisotropic hardening rule qualitatively gave the most accurate prediction in shapes of valley around both 0° and 90°. The reason of such a high accuracy in this condition can be that the influence of the shear stress is appropriately reflected.

In the present study, we have elucidated the behavior of a suspension composed of magnetic disk-like particles that flow in a Hagen-Poiseuille flow in a gradient magnetic field and the trapping characteristics of the magnetic particles by means of multi-pairs of magnetic poles. Brownian dynamics simulations were performed in order to clarify the dependence of the trapping characteristics on a variety of factors such as the magnetic interaction between particles, the strength of a non-uniform magnetic field, the strength of the flow field and the separation distance between two pairs of magnetic poles. The main results obtained here are summarized as follows. An increase in the magnetic field strength improves the trapping characteristics of disk-like particles if the influence of an applied magnetic field is sufficiently more dominant than that of a flow field and also that of the magnetic particle-particle interaction. In the case of a strong magnetic particle-particle interaction, thin chain-like clusters are formed from an anchored particle trapped around a magnetic pole. Moreover, if the separation distance between the two poles is sufficiently near, an arch-type cluster is formed between the magnetic poles, and is located in a vicinity area nearer to the wall surface due to the influence of a flow field. For the case of a small aspect ratio r_p=3, in the situation that the particle volumetric fraction is constant, an arch-type cluster formed between the magnetic poles can remain trapped in the vicinity of the wall surface even under the influence of the strong flow field. Hence, we understand that the use of disk-like particles with a small aspect ratio may be able to improve the trapping performance of particles.

The equation model of disintegration and agglomeration has been developed to analyze the performance of the bead mill. The equation model, derived under the assumption that the frequency of disintegration depends on the volume ratio of the agglomerates, comprises two parameters: the disintegration coefficient and the agglomeration coefficient. Six experiments were conducted to determine these two coefficients. When a common value is applied for the agglomeration coefficient, the value of the disintegration coefficient reflects the performance of the bead mill. This proposed method could assist in optimizing the conditions for bead mills.

An Excel program that uses tri-diagonal matrix algorithm (TDMA) is proposed to calculate one dimensional steady heat conduction. The program gives analytical exact solutions to calculate multi-layered symmetric flat plate, solid cylinder, sphere, double-sided flat plate, hollow cylinder, and spherical shell geometries. Three different boundary conditions and arbitrary heat generation rates can be applied. Most existing numerical analysis methods use the finite difference approximation. When the thickness and physical properties of adjacent layers are extremely different in the vicinity layers, accurate values cannot be obtained without a very large number of calculation meshes. A large number of meshes increases calculation time, even for simple geometries. However, the analytical method described in this paper does not affect the accuracy of the analysis even if the adjacent media have extremely different thicknesses, because an exact solution is obtained. The effectiveness of the program is demonstrated by solving an example using this Excel program. Furthermore, the method is applied various engineering problems to show how to make simplified heat transfer models to the realistic engineering problems. By using this program to perform heat transfer calculation examples described in JIS, laws and regulations, it was easy to confirm the validity of the regulations and standard settings. The excel program is open to the public for engineering design.

This paper proposes a novel path planning algorithm designed for efficient lunar water exploration. The proposed algorithm leverages a Bayesian network-based sensor model in combination with the RRT* algorithm. It is well-known that water is highly likely to be present in permanently shadowed regions (PSRs). However, during exploration within these regions, the rover must escape to illuminated areas within a limited time to recharge its battery. Therefore, it is crucial to develop a path planning algorithm that ensures both timely battery recharging and efficient water exploration. To address this challenge, we first introduce a sensor model based on a Bayesian network for the neutron spectrometer. Subsequently, we propose an algorithm that utilizes this sensor model and the RRT* algorithm to effectively explore water on the lunar surface. The paper provides a numerical simulation, using real data about time-variant illumination areas and time-invariant PSRs on the lunar surface, to illustrate the effectiveness of the proposed algorithm. The results reveal that the algorithm accurately and efficiently detects water under realistic illumination conditions.

In noisy environments such as factories or construction sites, noise should be reduced and the speech of workers moving in the environments should be recognized. Noise control is a method for reducing existing sounds by producing a secondary sound with the same amplitude but with antiphase, and effective on low frequency noise reduction. However, conventional noise control systems reduce all existing sounds but cannot preserve a specific target sound such as speech in noise. In this study, we propose a novel system which selectively reduces noise and preserves target sound by combining a noise control system with a sound source separation system using multilayer neural networks. The sound source separation part of the system was designed to separate mixture into noise for reduction and target sound for preservation and the use as a teaching signal of supervised learning of the neural networks. The noise control part of the system was designed to reduce noise at fixed and moving evaluation points. The whole system was designed to realize high noise reduction performance preserving target sound and low computational cost for real-time control in experiments. The performance of the proposed system was experimentally verified under the conditions that the noise and target waveforms were set as both sine waves, both superposed sine waves, and band noise and a sine wave, respectively. The results indicate that the proposed system preserves target sound to some extent and reduces noise by more than 10 dB and approximately 5–6 dB at the fixed and moving evaluation points, respectively.

We investigated the damping mechanism of granular material dampers when using a structure with a relatively high natural frequency and small vibration displacement as the target of vibration damping. For powder or granular material dampers, the movement of the powder or granular material is the basic principle of damping. Damping ratios were calculated and compared by individually modifying each parameter (other than diameter) that affects the granule’s motion. As a result, it was found that when the moment of inertia of the granules is 1/2 and the Young's modulus is 1/100 of steel, the damping effect is improved in the region where the total body weight is heavy. In addition, as a specific example of reducing the moment of inertia and optimizing Young’s modulus, when the effect was examined by calculation and experiment using rubber ball with steel core, it was found that the damping ratio can be improved in the region where the total mass of the granules is large. Furthermore, we obtained new knowledge that the hardening spring characteristics of rubber balls containing steel balls have a positive effect on damping characteristics.

With the recent spread of 3D printing, its application has been widely discussed from automotive industries to small-scale and individual manufacturing. In general, additive manufacturing is used as a 3D printing process. Using this process, defects in one layer can easily spread to the next layer, deteriorating the overall quality of the final product. Therefore, to ensure the quality of 3D printers while maintaining a large workspace, it is important to monitor the process online. In this paper, a coarse-to-fine strategy is implemented for Fused Deposition Modeling via the dynamic compensation module attached to an industrial robot arm, and the 3D printer-specific multi-ROIs image processing method is proposed for printing monitoring and compensation. The dynamic compensation module consists of two high-speed cameras, an extruder, and a 2-axis actuator that can monitor previously printed material to compensate for the movement of the main system as a high-frequency sub-system. For evaluation of the developed system, a defection detecting test using the multi-ROIs method and dynamic compensation tool path planning test was made. Consequently, it was shown that the dynamic compensation system can be applied by dynamic compensation of the developed 3D printer system, and the prediction and inspection can be made simultaneously by using the multi-ROIs method.

In the construction work, flat plates for building materials are transported by a suspension mechanism using wires in air flow. Then, a flutter could be caused to the suspended plates due to the interaction between plate motion and fluid flow. The violent vibration due to flutter loses the efficiency and safety of the operation. To improve the efficiency and safety of the operation, a detailed understanding of the flutter characteristics is crucial. In this paper, the flutter analysis and experiments of a flat plate suspended by wires in air flow are carried out. In the flutter analysis, the Doublet-point method based on the unsteady lifting surface theory is used to calculate the unsteady fluid force acting on the plate surface. The equation of motion of the plate suspended by wires considering restoring force due to gravity is derived based on the Lagrange equation. The flat plate is modeled as a rigid body. The flutter velocity and frequency are examined through the root locus of the characteristic equation of the system with changing the flow velocity. Moreover, wind - tunnel experiments are conducted to verify the analytical results. The influence of a suspending angle on the flutter velocity, frequency and mode are investigated in detail. Finally, the work done by the fluid force acting on the plate surface is examined.

In recent years, the research on airless tire (Non-pneumatic tire, NPT) for passenger cars has received extensive attention. The background to this is the development of autonomous driving technology. Autonomous driving of Level 4 and above is un-manned driving, so it is more important to reduce the frequency of maintenance due to accidental failures, and punctureless airless tires are expected as tires for autonomous driving vehicles. Rolling resistance of tires is a highly important property for tire development. The reduction of rolling resistance of tires has become more significant in the tire industry, because it has close relation to the fuel consumption efficiency. As for conventional pneumatic tire, the rolling resistance is mainly generated by energy loss of tire components during tire deformation and it is well known that to reduce the rolling resistance, tanδ of rubber should be decreased. The purpose of this study is to clarify the contributing factors of airless tires to rolling resistance. As a result of measuring the tire force generated by a rolling experiment using a scale model airless tire, it was confirmed that the wheel load fluctuated impulsively. Therefore, we observed the tire deformation during rolling with a high-speed video camera and clarified the mechanism of tire force generation.

In this study, we proposed a structural health monitoring and diagnostic method for layered structures using the Autoencoder (AE). This method belongs to the primary diagnosis one, and its purpose is to identify the location of abnormality quickly after abnormality detection. An abnormality represents a decrease in the stiffness characteristics (spring constant) of the outer wall of a hierarchical structure when it deteriorates or is damaged. The proposed method has the following two features. The one is that the AE learns only the Frequency Response Function (FRF) under normal conditions calculated by the mathematical model of the layered structure. The FRFs of the actual structure under abnormal conditions are not required for the learning data. The other is that the FRFs under abnormal condition computed by the mathematical model are input to the learned AE to obtain a standard of the initial abnormality. The location of abnormality is identified by checking that the abnormality in the actual structure fits the standard of the initial abnormality. First, we considered a three-layered structure as a numerical example and verified the validity of the proposed method. When the method was applied to the three types of abnormal conditions, it was shown that the abnormal diagnosis could be performed correctly. Next, we constructed an experimental model of a three-layered structure, and realized three types of abnormal states similar to the numerical examples. We verified the applicability of the proposed method and showed that correct abnormal diagnosis was possible. As described above, the validity and applicability of the proposed method were clarified.

Overhead crane is a product used to transport heavy objects in a factory, and advanced operating skill is required to operate safely and with less equipment failure. However, the shortage of skilled operators has been pointed out in recent years. The purpose of this paper is to develop a method for estimating tense state of rope and weight of suspended load at the same time with the aim of improving the safety of overhead cranes without relying on operator skills. When estimating multiple states using independent estimation models, the issue is the lack of computer storage space and calculation speed due to the increase of the parameters. Therefore, the authors adopted a method that simultaneously estimates several states in a single model. To estimate the tense state of rope and weight of suspended load from the driving signals of the induction motor mounted on the overhead crane, a multi-task neural network that can express the nonlinearity of motor characteristic was used. The teaching data used for learning network parameters was created by using object detection technology to quantify the transportation of markers attached on load block and suspended load. As a result, it was clarified that the rope tension state and the load can be estimated simultaneously by inputting the driving signals of the motor to the multi-task neural network. Moreover, it was shown that the rope tension probability and the estimated load increased synchronously as the motor signals change, and the tension detection was output.

The study proposes a supporting method in the upstream design process, utilizing the Top Trading Cycles (TTC) algorithm in matching theory. A decision in upstream design is important because any decision in this process strongly affects the final product’s architecture or quality. However, decisions in the design process are subjectively conducted by designers, which results in a lack of objectivity and reproducibility. To tackle this issue, the present study models a matching problem between a function element set and a value element set. After the mathematical formulation, a new matching algorithm is proposed. In the algorithm, mutual/self cycles between function elements and value elements are prioritized to be matched, and furthermore, matches can be resolved and rematched with a new entity in order to avoid forming blocking pairs. As a result, stable matching is able to be attained, whatever preferences are. The difference from the normal TTC is that the proposed matching algorithm can realize stability even if indifferent preferences are included. Then, a case study was conducted where the proposed method was applied to the issue of function selection about car navigation systems. Finally, the effectiveness is discussed from various points of view. The proposed matching algorithm between function elements and value elements could potentially contribute to objectivity and reproducibility in an upstream design phase.

In order to obtain the fundamental information for the measurement of thin oil film thickness adhered on lubrication surface using ultrasonic method, the ultrasonic calculation considering the scattering attenuation at each interface and the influence of frequency spectrum of ultrasonic waves were performed. The echo ratio of the wave reflected from the adhered film is hardly affected by the roughness on general lubrication surfaces with R_q≦1μm. On the other hand, the echo ratio is strongly affected by the frequency spectrum, and the relationship between the adhered film thickness and the echo ratio (calibration curve) calculated assuming the center frequency and frequency spectrum of the measured wave almost agrees with the experimental results. Based on the echo ratio distribution measured by the cylinder sliding type oil film thickness observation device, adhered film thickness distribution of the engine cylinder was plotted. Adhered oil film thickness increases and decreases symmetrically on the inlet and outlet sides of the piston ring when the oil supply is sufficient. However, the thickness of the adhered oil film becomes thin and its distribution is flattened throughout the stroke when the oil flow rate is reduced. On the other hand, the relationship between the film thicknesses L_A on the cylinder and the film thickness L_R on the piston ring for a sufficient oil flow rate is close to the shape of infinity (∞), and adhered film thicknesses of the lower side of top ring (L_AT) and of the upper side of second ring (L_AS) show a negative correlation. Therefore, these correlations of film thickness behavior have a potential to be one of the indicators of deterioration of lubrication conditions.

Production system simulation is a powerful tool to achieve efficient operations in complicated production systems such as high-mix and low-volume production. However, it takes significant efforts and expertise to construct accurate simulation models. In this article, a novel modeling approach called as data-driven and multi-scale modeling is proposed. The proposed approach combines various modeling methods to maximize the simulation accuracy. In order to verify the usefulness of the proposed approach, computational experiments for simple production systems to compare modeling methods are conducted. The experimental results show that the superiority of modeling methods depends on the background knowledge and available information about target production system and the proper use of modeling methods is important to achieve high accuracy.

Walking is one of the most important activities. However, when abnormalities occur during walking, falls occur. Fractures caused by falls may significantly reduce activities of daily living and quality of life. Falls due to external factors occur during the swing phase when the leg is floating. Detection of the period from the moment the lower leg becomes vertical to the moment of the heel contact may help elucidate the phenomenon of falls in the future and may prevent falls by some manipulation during this period. Therefore, in this study, we develop an algorithm for detecting terminal swing independent of walking speed using a 6-axis sensor attached to the heel of shoes. Preliminary experiments were conducted on a treadmill at walking speeds of 2, 3, 4, and 5 km/h. The theoretical time of the terminal swing was measured using motion capture. Preliminary experimental results showed a correlation between the maximum angular velocity of the lateral direction axis of the 6-axis sensor and the acceleration in the toe direction at the beginning of the terminal swing. Therefore, this study proposes an algorithm for detecting terminal swing using the relationship between the maximum angular velocity of the lateral direction axis and the acceleration in the toe direction at the beginning of the terminal swing. Comparing the time of the beginning of the terminal swing between the proposed method and motion capture, it was confirmed that the algorithm could detect the terminal swing at walking speeds of 3, 4, and 5 km/h with a smaller error than in previous studies.