In ball-mill grinding, the internal state of comminution is difficult to monitor because the mill is enclosed, and stop timing often depends on operator experience. This study investigates the analytical conditions for unsupervised discrimination of particle size using radiated sound from an operating laboratory-scale ball mill. The radiated sound pressure was measured with four microphones placed around the mill wall, and frequency spectra were obtained. The obtained spectra were utilized as input features of a one-dimensional self-organizing map (SOM) to cluster acoustic patterns without prior training data on particle-size distributions. To assign a physical scale to the SOM units, they were rearranged according to the average overall value (OA) of the spectrum classified into each unit. The discrimination performance was evaluated using the mean unit index for six particle sizes ranging from 0.5 to 10 mm under identical operating conditions. Furthermore, the effects of the analyzed frequency band and FFT time-window length were investigated. The high-frequency band reduced the variance of unit indices and improved the separability among particle sizes, whereas a 0.2 s window with 50% overlap achieved the best balance between maintaining separability and securing a sufficient number of spectra. These results provide practical guidelines for applying SOM-based unsupervised monitoring to in-situ estimation of grinding progress.
The rapid proliferation of unmanned aerial vehicles (drones) has made ensuring safety through non-cooperative collision avoidance an urgent priority. Acoustic sensing offers a lightweight solution; however, detecting approaching rotorcraft, such as helicopters, remains challenging due to the high levels of drone self-noise. To address this challenge, this study proposes a robust sound source detection and direction estimation method focusing on the amplitude modulation (AM) characteristics inherent in rotorcraft noise. First, utilizing the cyclostationary nature of rotorcraft noise, we adopted the Fast-Spectral Correlation (Fast-SC) algorithm. By calculating the Enhanced Envelope Spectrum (EES), we successfully separated helicopter signals, which exhibit specific low-frequency modulation, from drone noise containing high-frequency modulation. Second, we developed a direction estimation technique integrating this AM analysis with a fixed delay-and-sum beamformer (DSBF) using a linear microphone array. By scanning the beam and evaluating the spatial distribution of AM intensity, the system can accurately estimate the direction of the approaching helicopter. Furthermore, we conducted a comprehensive investigation to optimize the microphone array configuration, specifically analyzing the effects of the number of microphone elements and their spacing on AM directional characteristics. The results from numerical simulations and field verification experiments showed high consistency. Consequently, suitable design parameters maximizing detection performance were identified, demonstrating the applicability of the system to practical collision avoidance.
This paper investigates structural improvements for a passive landing shock-response control mechanism for aerial transport devices that combines a support spring with a granular damper. The primary objective is to suppress post-landing rebound to avoid attitude disturbance and mission interruption, while maintaining the peak acceleration at touchdown within a certain range without external power or active control. In the proposed mechanism, the impact energy is temporarily stored in the spring and then transmitted to the particle bed through a compression plate, and dissipated mainly through repeated particle–particle and particle–wall collisions, frictional sliding, and material losses. To clarify design effects on rebound suppression, spring stiffness and compression-plate geometry are selected as design factors and systematically varied under identical landing conditions. Numerical analyses are performed using a discrete element method coupled with rigid-body dynamics, with a contact model appropriate for the particle material. The results show that lower spring stiffness consistently reduces the maximum rebound height and also tends to mitigate the acceleration response immediately after touchdown. Moreover, compression-plate geometry significantly affects rebound suppression by altering particle rearrangement and compaction behavior. The peak acceleration at touchdown is within a certain range in most cases. These findings provide useful design guidelines for landing response control.
In the present study, we investigated the transition in aggregate structures of a suspension composed of magnetic Janus particles under a uniform magnetic field. Monte Carlo simulations were performed to investigate how the transition of aggregate structures depends on various factors, such as the magnetic interaction between particles and the strength of an applied magnetic field. Here, we employ a particle model that has a magnetic moment radially shifted with respect to the particle center. The main results obtained here are summarized as follows. Janus particles with magnetic moments near their centers tend to aggregate and form chain-like clusters as the magnetic interaction strength increases. In the case of strong magnetic interactions, an increase in the distance between the magnetic moment and the particle center leads to a transition in aggregate structures from chain-like clusters to packed clusters. These cluster units tend to consist of 2 to 5 particles. As the magnetic field strength increases, chain-like clusters tend to be inclined toward the magnetic field direction. For Janus particles with magnetic moments near their surfaces, constituent particles in chain-like clusters tend to arrange so that their magnetic caps are close to each other. From these results, it is evident that both the strength of magnetic particle-particle interaction and the applied magnetic field strength have a significant influence on the internal structure of Janus particle aggregates.
This paper proposes an efficient method for evaluating the intensity of the singular stress field (ISSF) at an interface corner in a three-dimensional (3D) bonded structure, where two distinct real singularity indices are involved. Finite element method (FEM) analyses are conducted for both unknown and reference problems using the similar mesh patterns. A known relation exists between the FEM-derived stress σFEM, the minimum element size emin and the singularity index λ as σFEM ·(emin)1−λ= constant. Based on this relation, two singularity indices are simultaneously determined, and the FEM stress is discomposed into two components corresponding to each index. Then, the ISSF for the unknown problem is obtained from the FEM stress ratio of the reference and unknown problems. To demonstrate effectiveness of the proposed method, it is applied to a 3D scarf joint with a 45-degree scarf angle composed of steel adherend and epoxy adhesive. The results show that the ISSF can be analyzed with accuracy comparable to that of the H-integral method. Moreover, the 3D corner ISSF for the scarf joint is analyzed by varying width W, depth D and adhesive layer thickness h. It is found that when h/min(D,W) ≤ 0.1, the 3D corner ISSF depends solely on h. It is confirmed that the same failure stress can be obtained when the adhesive layer satisfies the condition and has the same thickness by using the previous experiment result.
eHMI (external Human Machine Interface) that displays texts and signs as a method of conveying a vehicle’s guidance intention into pedestrians has been actively developed. However, current eHMI has problems where the vehicle’s intention and purpose are difficult to understand for pedestrians, or the differences in language and culture lead to various interpretations. Therefore, a method of conveying intentions that pedestrians intuitively understand the vehicle’s intentions is required. This study focused on the eyes, which facilitate shared intentions in human-human communication, and developed an eye-type eHMI “i-EyFuze.” i-EyFuze has an intention conveying method that guides pedestrians by extracting the target pedestrian through image processing and superimposing the image of pedestrian on the vehicle’s gaze. It was confirmed that a self-shadow representation that protects interpersonal privacy is preferred for the image of superimposition. Furthermore, the evaluation of impressions of i-EyFuze based on a simulating traffic environment that mimics a pedestrian crossing confirmed the effectiveness of the proposed self-shadow. In addition, the evaluation of intention conveyance in the same environment showed that the conveyance of pedestrian guidance suggests may be enhanced when the moving self-shadow involves a slightly delay for vehicle’s gaze.
Recent regulations on the spring-like effect (SLE) in golf drivers cap the characteristic time (CT) at 239±18 μs, which limits performance gains derived solely from face deflection. This study investigates an alternative approach: the design of sole geometries inspired by origami engineering. Building on folding patterns such as Miura-ori and Yoshimura, we developed analytical models incorporating origami structures into the sole. Finite element frequency-response analyses were conducted to compare face amplitudes, revealing that the application of a K-fold configuration significantly increases face displacement compared to a baseline model. To quantitatively evaluate the influence of design factors—specifically the number of longitudinal and transverse fold lines, zigzag ratio, and fold depth—on face amplitude, a parametric study was performed. Optimal conditions for the target club geometry were identified as (m, n, s, h) = (7, 5, 1/2, 1.5). The optimized K-fold model exhibited approximately 18.3 times larger face deflection and 16.9 times larger sole deflection versus the baseline, demonstrating a substantial increase in global head compliance. Furthermore, a physical prototype was fabricated based on the analytical results, and robot hitting tests were performed to validate the simulation and evaluate the effectiveness of the K-fold structure. The results confirm that the K-fold application enhances the rebound performance by increasing ball launch speed and carry distance without altering the overall stiffness of the club. These findings establish that origami-derived architectures are an effective mechanism for improving ball-flight performance under existing rule constraints.
One key challenge in the system integration business is improving the efficiency of schema matching. Schema matching is a task where two different schemas are given as input, and semantically corresponding columns are identified. Recent methods using large language models (LLM) outperform rule-based and deep-learning approaches. However, the LLM-based approaches face two challenges: improving matching accuracy of (i) one-to-many cardinality and (ii) foreign keys (FKs). For the first challenge, we focused on the features that the one-to-many or one-to-zero cardinalities can be excluded if the column is an identifier (ID) or essential for the target application (i.e. production planning system), respectively, and thus the possible multiplicity can be classified into four patterns. Based on the features, we proposed a decomposition approach where an original problem is divided into four subproblems according to the four multiplicity patterns, and we use prompts specialized for each subproblem. For the second challenge, we focused on the features that database schemas can be modeled as graphs, and the FKs can be identified in the shortest path between the primary keys (PKs); thus, FKs can be identified by graph path search between PKs. Based on the features, we proposed a graph-based approach where PKs are matched first using LLM, and each FK is matched later using graph path search between corresponding PKs. The effectiveness of our proposed approaches was demonstrated through an industrial case study.
This study developed a pneumatic adaptive securing band (PneumaBand), which can gently hold large objects of various shapes using an inverse pneumatic artificial muscle (IPAM). An IPAM is a soft actuator that elongates when its internal air pressure increases and contracts when the internal air is evacuated. If PneumaBand is pre‑pressurized, it can be activated simply by releasing the internal pressure to the atmosphere, requiring no external power source during operation. It can also be used with a compressed gas cylinder or a compact air pump. PneumaBand can expand its inner diameter up to three times its original size and hold large objects with a circumference of exceeding one meter. It is foldable for compact storage, allows adjustment of the wrapping force through internal air pressure or manual changes in its initial size, features a simple and durable structure, and is suitable for outdoor use. In addition, it retains its wrapping force even if damaged. This paper describes the mechanical properties of the IPAM as well as the wrapping force and wrapping speed of PneumaBand, and discusses potential applications based on experiments. Experimental results show that PneumaBand can be used to secure and protect objects, fix items on shipping pallets, hang large objects, and function as a robot gripper.
Active magnetic bearings (AMBs) have generally been implemented on machines on rigid bases. However, when integrated into ships or vehicles, base excitation poses a significant challenge due to the inherently low stiffness of AMB systems. This study aims to reduce the vibration response of levitated rotors subjected to pitching base excitation. The response characteristics of both non-rotating and rotating rotors connected via spring couplings were experimentally investigated under vertical and pitching excitations. A feedforward control approach using accelerometer signals was implemented to estimate and cancel inertial forces arising from base motion. Under vertical harmonic excitation (10 Hz, 6.3 m/s²), the proposed method reduced rotor vibration amplitudes by more than 64%, successfully preventing contact with the touch-down bearings. Under pitching excitation (10 Hz, 15 rad/s²), which induced vibrations in both the radial and axial directions, the feedforward controller effectively reduced vibrations in both directions.
Accurate identification of hysteretic restoring‑force parameters in civil structures usually requires displacement or strain measurements that are rarely available in field monitoring. This study proposes a sequential Bayesian updating framework that estimates confidence intervals for seven hysteresis parameters and four algorithmic hyperparameters (11 variables in total) solely from absolute acceleration records. Initial particle populations are generated by Evolutionary Computation, and the posterior distribution is progressively refined with a Sequential Monte Carlo sampler in which a kernel‑density‑estimated likelihood accommodates multi‑modal residuals. The approach eliminates manual tuning of algorithmic hyper‑parameters by learning the KDE (Kernel Density Estimation) bandwidth and PSO (Particle Swarm Optimization) coefficients simultaneously with the physical parameters. Numerical verification using a single‑degree‑of‑freedom system with a bilinear skeleton curve lacking point symmetry about the origin demonstrates that, after assimilating 100 % of the simulated earthquake data, the posterior standard deviations shrink by more than 70 % while the 95 % confidence intervals capture the true values of all variables. Throughout the three data‑assimilation stages, the effective sample size remains above 90 %, indicating negligible particle degeneracy.
