-
Yuya Oda, Kazuya OIZUMI
Session ID: 2208
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
For the manufacturing industry, comunication among designers is becoming more difficult due to the increasing scale and complexity of products. While the results of designs are rather easy to be communicated as data, the design rationales, which is rarely depicted in data, would be difficult to be communicated. The purpose of this sutudy is to store rationales while a product is designed with a low workload. It is exprctrd that there are changes in designers’ states when designers ideate something. This paper proposes a method to automatically extract design rationales using 2D pose data obtained withOpenPose, which can be used to estimate states of designers. . As a result of the validation of the method, it was possible to extract design rationales with relatively high accuracy, under the condition that poses to be observed would be selected differently to each designer.
View full abstract
-
Motofumi IWATA, Kohei SHINTANI, Ryuichi NODA, Kizuku MINETA, Makoto MU ...
Session ID: 2301
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
-
Sakura KAWATA, Kohsuke MIURA, Kazuyuki HANAHARA, Kaori YAMADA
Session ID: 2302
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Underwater robots typically employ propellers as their propulsion system; they also have steering mechanisms and additional actuation devices to be used for such as maintaining their intended posture. On the other hand, many of the aerial drones are equipped only with four propellers; still they have enough ability to freely fly in the air. Suppressing the number of propellers also contributes to their cost-effectiveness. In this report, we present a design of a quad-thruster-type underwater drone that has a corresponding movement capabilities to the aerial drones. The proposed underwater drone is capable of maintaining its posture in motion in every direction: forward, backward, left, right, upward, and downward. Unlike conventional underwater robots relying on steering mechanisms, this drone takes advantage of the negative buoyancy in order to attain such motion capabilities. This study involves a fundamental investigation into the stable translational movement of the proposed underwater vehicle, focusing on posture control such as hovering behavior. The conceptual design, the embodiment design and the detailed design of the proposed underwater vehicle are explained. This study will also introduce the characteristics of the proposed ROV design and innovative ideas based on preliminary experiments.
View full abstract
-
Ryusho KITAGAWA, Yosuke KADOKAWA, Toshiki SIMA, Yuki SHIMIZUGUCHI
Session ID: 2303
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
-
Changyoung YUHN, Yuki SATO, Hiroki KOBAYASHI, Atsushi KAWAMOTO, Tsuyos ...
Session ID: 2305
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
-
Kenta KOYAMA, Nozomu KOGISO
Session ID: 2306
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
-
Daiki MORITA, Saki INAGAKI, Yoshihiro SEJIMA
Session ID: 2308
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
As the saying goes, “Two minds are better than one”, group communication can generate more ideas than individual thinking. To generate good ideas, a good relationship in a group as well as an activated and smooth interaction are essential during group communication. Previous research proposed many idea generation systems that are intervened directly in the idea during group communication. However, an approach to indirectly support idea generation by enhancing group communication has not been sufficiently examined. In this study, we constructed a system that visually presents the activation level in group communication. This system is projected the Ebbinghaus illusion on a desk and designed to change it based on the activation level that estimated using our group-activated communication model. The effectiveness of the system demonstrated by sensory evaluations.
View full abstract
-
Kento SHIMANUKI, Yoshikatsu KISANUKI, Hiromitu SUZUKI, Hideki SUGIURA
Session ID: 2309
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
-
(Development of a small racing car considering resource management and manufacturability/workability)
Kei KUROISHI, Syota TATEYAMA, Ryusaku NOGUCHI, Ryuto TAKAOKA, Atsuhiro ...
Session ID: 2311
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
-
Naoya MATSUMURA, Ryosuke SAWAZUMI, Takuya SUGIURA
Session ID: 2312
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In the early stages of the design process for automotive components, it is important to look over various design plans and select a geometry with good tendency. Currently, CAE (Computer Aided Engineering) is used to predict the performance of products, but it takes time to evaluate the performance of a single geometry, and it is necessary to repeate change of the geometry and numeric calculation to achieve the target performance. In this study, a convolutional neural network, a machine learning method for image recognition, is used to construct a system that can quickly predict the crash performance of a product by relating shape information to CAE results.
View full abstract
-
Misato KATO, Taisei KII, Kentaro YAJI, Kikuo FUJITA
Session ID: 2314
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
From the perspective of structural design, maximum stress minimization is an active research on topology optimization. While gradient-based topology optimization is used in most conventional methods, its formulation requires several approximations and relaxation techniques. This has resulted in a discrepancy between the optimized results involving relaxation techniques and the analytical results obtained from the original problem. Additionally, in the conventional methods, stress concentration occurs at stair-stepped boundaries due to the use of structured mesh. Therefore, analyzing on body-fitted mesh is essential for accurately evaluating the performance of the optimized results. In this study, we focus on data-driven multifidelity topology design (MFTD) and propose a framework for optimization based on higherprecision analytical results obtained from the original problem using body-fitted mesh. In the numerical examples, we tackle a bi-objective problem of the exact maximum stress and volume minimization. We first obtain solutions through gradient-based topology optimization using the conventional density approach. Subsequently, we use these solutions as the initial solutions and validate whether the performance of these solutions improves by our proposed method. The numerical results indicate that the optimized results by data-driven MFTD completely surpassed the performance of the initial solutions given by solving gradient-based topology optimization.
View full abstract
-
Kotaro TAKENAKA, Makoto YAMAKAWA, Kazuki HAYASHI
Session ID: 2315
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
With the recent development of machine learning methods, the application of machine learning to the optimal design of building structures has been considered. Among them, the effectiveness of reinforcement learning for planar frames has been reported. However, direct application to structures with a large number of members and cross-sections, such as space frames, remains a challenge. In this study, we propose a method using reinforcement learning based on a policy network for the problem of minimizing the amount of steel members in a space steel frame with discrete cross-sectional specifications by decomposing and reconstructing it into a planar frame. The reinforcement learning agent learns to select appropriate actions for members that violate constraints, thereby acquiring the ability to reduce the amount of steel while satisfying constraints from the initial cross-section design.
View full abstract
-
Ryohei HAZAMA, Nami OKAMURA, Nurhayati Md Issa, Wira Jazair Yahya, Moh ...
Session ID: 2316
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Latent needs are important in product development because they represent unarticulated customer desires that, when identified and addressed, can lead to innovative and successful products that meet evolving market demands. In our research, various kinds of co-creation ideas will be stimulated, and latent needs could be extracted if a working prototype is introduced directly to the potential user was assumed. The interviews were conducted after showing them the working prototype. The interview responses were then interpreted using the interpretation guideline commonly used by other designers. Then, new additional guidelines were introduced after considering that experience, empathy, and knowledge of working prototypes are important and able to extract more latent needs. The new guidelines are "to write a statement with empathy", "to write a statement as a designer", and "to write a statement as someone with experience". From the research result, it is concluded that the number of interpreted needs increased by a small amount when the new proposed guidelines were additionally applied. The results also indicated that the needs interpreted with new guidelines obtain high scores with our quantitative interpreted needs evaluation method. The application of the proposed guidelines showed the possibility of discovering needs by considering empathy, experience, and knowledge of working prototypes.
View full abstract
-
Nurhayati Md ISSA, Nami OKAMURA, Ryohei HAZAMA, Wira Jazair YAHYA, Moh ...
Session ID: 2317
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Reflecting latent needs in product functionality is an important step in the product development process. "Those that many customers identify as significant but cannot be adequately articulated in advance" is how latent demands are described. In addition, it is difficult to extract latent needs through quantitative analysis. In this study, a working prototype is introduced and interviews are conducted to verify the method of eliciting latent needs. The results of introducing a working prototype focusing on the "preventing entry to dangerous areas" function are presented. Then, a method to examine the effectiveness of the quantitative evaluation method was evaluated by applying the Anderson-Darling test, and it was shown that the method follows a normal distribution for our interpreted needs with the comprehensive working prototype. Therefore, it can be said that this evaluation method can be applied to the assessment of important latent needs.
View full abstract
-
Ryota SAWAUCHI, Takashi OYAMA, Teruaki ITO
Session ID: 2318
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Various methods have been explored for social media text analysis, including sentiment classification, spam detection, and identifying inflammatory videos. However, the format of free text makes it difficult to obtain consistent analysis results. To address this, sentiment-based analysis and classification are conducted. The accuracy of analyzing videos using multi-class sentiment estimation is considered lower compared to using negative and positive sentiment models. To improve accuracy, the WRIME dataset is used, and an emotion intensity estimation model is generated by retraining a pre-trained BERT linguistic model. The study aims to extract objective features from subjective free text data using machine learning and provide classifications and discussions. The experiment focuses on comments from Vocaloid producers' YouTube videos, and the results indicate the effectiveness of 8 emotion estimation for video analysis, suggesting potential applications in emotion-aware recommendation and analytics systems.
View full abstract
-
Yudai UMEZAWA, Naoyuki ISHIDA, Hao LI, Tsuguo KONDO, Kozo FURUTA, Sung ...
Session ID: 2401
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In October 2020, Japan declared its goal of becoming a carbon-neutral society by 2050 in response to the worsening global warming(1). In the aviation industry, the number of passengers is expected to increase in the future(2), and we have to develop high-performance wings to improve fuel efficiency. In recent years, Computational Fluid Dynamics (CFD) has been widely used to design aircraft wings, and numerous optimization methods have been proposed. Above all, topology optimization stands out as the most flexible design approach and allows changes in morphology that generate holes in the interior of material parts (3). For application to aircraft design, Kondo et al. (4) applied topology optimization to drag minimization and lift maximization problems. They successfully applied topology optimization to minimize drag and maximize lift under volume constraints. However, although it is important to consider the projected area in aircraft design, no drag minimization or lift maximization problems have been formulated with the projected area as a constraint. In this study, a drag minimization problem under the projected area constraint based on topology optimization is formulated by introducing a virtual physical field. As a result, we obtained shapes that differed from those achieved under volume constraints. In this study, the method is based on using the advection-diffusion equation to express the projected area constraint. Topology optimization was performed using the level set method and updated using the reaction-diffusion equation. The software used for the calculations was FREEFEM++.
View full abstract
-
Rikuto YAMADA, Jin-Xing SHI
Session ID: 2402
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Lure fishing is a special type of game fishing using artificial fishing lures instead of live baits and has been widely enjoyed around the world. Soft lures with jig heads are a kind of fishing lures that perform vibration behavior by the soft body to attract carnivorous fish. It can be assumed that enhancing their vibration behavior can attract carnivorous fish efficiently. In general, lure design is significantly based on the intuition of the designer and only a few academic works focus on it. In this work, we aim to perform structural design optimization of jig-head soft lures using fluid–structure interaction (FSI) analysis and response surface methodology. At first, we construct the analytical model of a commercial jig-head soft lure for FSI analysis by using a 3D scanner. Then, we perform FSI analysis of the jig-head soft lure using the finite element method (FEM) to reproduce its underwater motion of 3 seconds and obtain the displacements of a designated node in the soft body. According to the results of displacements, we calculate the frequency within the 2s ~ 3s underwater motion using the fast Fourier transform (FFT), and Amount of variation in Y direction. Using the same procedure, we carry out 9 FSI analyses to calculate the frequencies by setting two design variables of the lure, which are the Young’s modulus of the jig head and the length of the soft body. At last, we adopt the results in the central composite design that belongs to the response surface methodology to obtain the optimal the vibration times in 3 seconds, which reaches to its highest value of 11.68.
View full abstract
-
Yoshihiro KANNO
Session ID: 2408
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Data-driven approaches have recently achieved significant attention in diverse fields of science and engineering. Most of existing methods for data-driven computation in elasticity do not take uncertainty in the material constitutive law into account. This study presents an optimization-based method for finding a bound for the quality of the interest in static structural analysis of a linear elastic structure. It is guaranteed that, with the specified confidence level, the probability that the quality of interest is within the obtained bound is no smaller than the specified reliability. This theoretical guarantee is given by the fundamental property of the order statistics. It is shown that the bound is obtained by solving linear programming problems. We present a numerical example to show variations of the obtained bound with respect to the reliability and confidence levels.
View full abstract
-
Ryo TSUMOTO, Kazuya URATA, Yutaka NOMAGUCHI, Kentaro YAJI, Kikuo FUJIT ...
Session ID: 2409
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In structural design, the generative design that automatically generates many alternatives is gathering significant interest. It is because advanced topology optimization techniques that can determine an optimized material distribution within a given design domain. Its applications allow computers to generate novel and high-performance alternatives. On the other hand, generative design requires designers to explore various ones by interpreting their meanings as concepts. This paper first discusses the challenges of such concept identification from diverse alternatives produced by generative design. The discussion assumes that various structures can be categorized into several types, each of which is a material distribution under a high degree of design freedom, and recognizes a concept as a set of labels, each of which distinct a set of alternatives into different subsets, over the design representation. This paper then revisits a proposed framework for concept identification based on deep clustering and logistic regression. Finally, an application to a conceptual design problem of a flow distributor ascertains its fundamental validity and leads to future perspectives.
View full abstract
-
Teppei KIKUCHI, Kento TANAKA, Masayuki NAKAMURA
Session ID: 2412
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
By appropriately determining each layer's refractive index and film thickness, optical multilayer films can control the light spectrum using the interference effect of light. Combining solar and thermal power generation is possible using a heliostat with an optical multi-layer film attached to a solar cell panel in a tower-concentrated solar power generation facility. This research aims to design a super multilayer film that can control wide-angle and wide-band wavelengths based on the spectral sensitivity characteristics of solar cells. It is formulated as an optimization problem that seeks the optimum values of parameters such as the reflection center wavelength of one structural unit, the number of layers, and the incident angle. Design results for the case where the super multilayer film is attached to the back surface of the cover glass of the solar cell are reported.
View full abstract
-
(Potential of Motion Design Based on Emotion Mechanics)
Shimon HONDA, Takeo KATO, Hideyoshi YANAGISAWA
Session ID: 2501
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Mechanical design undertakes the process of determining behavior and structure from the required functions. We define "motion design" as the planning of behavior and structure that considers the affective quality in addition to the original function that the machine should fulfill. In this research, we focus on pleasant feelings, such as interest and admiration, that are obtained when we understand the structure that brings about the motion. Based on emotion mechanics, which explains the variation of human emotions using the variational principle, we hypothesize that these pleasant emotions are caused by switching models (a framework of cognition), i.e., model selection. Based on the idea that pleasant emotions arise when cognitive load decreases, we hypothesized that interest and admiration are maximized when complex (hard to predict) movements are realized with a simple structure (low degrees of freedom). To test the hypothesis, we conducted an experiment using an animation of a link mechanism to obtain impressions of the motion before and after showing the structure. The results showed an interaction between unpredictability and degree of freedom in the evaluation of interest. We also confirmed the main effect that admirations increase when the degree of freedom is low.
View full abstract
-
Toward the Development of General Features to Formulate Beauty of Shape
Hidefumi MATSUNAGA, Hideyoshi YANAGISAWA, Takeo KATO
Session ID: 2502
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Beauty of shape is an important factor in the visual evaluation of industrial products. This study proposes two new methods to formulate aesthetic features for shape using variational free energy (VFE) and marginal free energy (MFE), representing the prediction errors calculated in the human brain. VFE and EFE are used as indexes of emotion arousal potentials to form aesthetic liking functions, i.e., Wundt curve. We examine the relationship between each free energy calculated for butterfly shapes (as a natural object) and aesthetic evaluations obtained from human participants. The experimental results support the correspondence between these free energies and the arousal potential. Furthermore, we consider that each free energy has different characteristics depending on the difference in the formulation process. VFE is the aesthetic feature based on frequency distributions of shape feature states (i.e., curvature), while MFE additionally considers transitional probabilities between neighbor feature states (i.e., angle). The result suggests that MFE explains better regarding aesthetic interest and attractiveness than VFE.
View full abstract
-
Validation and Practical Application of Mathematical Modelling of Sensitivity to Novel Colors
Yuichiro OHASHI, Hideyoshi YANAGISAWA
Session ID: 2503
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Novelty is essential to product design, and novel colors enhance product appeal. With regard to novelty in product design, Loewy proposed the MAYA principle, which states that acceptable novelty is a condition for attractive design. Based on Berlyne’s arousal potential formulated in information-theoretic free energy, we propose a mathematical model of acceptability to novel colors. Our analysis of the mathematical model led to two hypotheses: first, the tolerance for novelty expands as the variance of the Bayesian likelihood function in the model increases, and second, as the variance increases, the colors that reach the optimal arousal level are more novel. We tested these two hypotheses in subject experiments to validate the mathematical model. Applying the model, we developed a suggestion system for novel colors. In the system, contour lines of the same arousal potential levels are drawn on the color coordinate space. By utilizing this system, it is expected to suggest attractive colors with acceptable novelty levels.
View full abstract
-
About the Use of the JSME Specification Template File
Tomokazu TAKEUCHI, Hideyoshi YANAGISAWA
Session ID: 2504
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
A problem in hybrid communication is that the conversation proceeds only with the face-to-face participants due to the lack of the presence of remote participants. In our research, we aimed to improve the presence of remote participants by designing the motion of the avatar robot. We hypothesize that the richness of model evidence for remote participant cognition corresponds to the presence of remote participants. We analyzed the elements of the presence of remote participants and focused on the state of "imposing" as model evidence of the presence of remote participants. We have not found any research so far that mathematically discusses the state of being “imposing” and the behaviors related to that state. Therefore, we mathematically modeled the behavior of imposing speakers by active inference. In our model, the state of being "imposing" corresponds to precision of action policy in active inference. Based on the mathematical model, we designed the behavior of the avatar robot that changes according to the precision. We confirmed the possibility of improving the presence of remote participants by designing the motion of the avatar robot with active inference.
View full abstract
-
Masakazu KOBAYASHI, Katsuyuki KUME
Session ID: 2507
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Recently, deep learning-based image generation AI such as Stable Diffusion, Midjourney, and DALL-E have been released and have attracted much attention. These methods are based on a latent diffusion model, a type of diffusion model, and can generate images that match the textual representation, called a prompt, input by the user. We believe that the use of image-generating AI in aesthetic design is promising because it can handle textual representations as prompts and can fine-tune the generated image by changing the textual representation of the prompt or by adding textual representation to the prompt. Therefore, as the first step in our research, we verified whether an image generation AI can generate product images that provide the impressions represented by Kansei words from the prompts that include those Kansei words, which is the most fundamental function of aesthetic design methods.
View full abstract
-
Aika YUKIMOTO, Tomoki SHIIYA, Takeo KATO, Takuya KATASHIMA, Yuka ODA, ...
Session ID: 2508
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Humans perceive softness by integrating physical properties such as stiffness and viscoelasticity, but the mechanism of perception and its relationship with physical properties are not clear. It is expected that gaining knowledge of the mechanism of mechanisms of tactile perception will help to create sensory value. In this study, we used Tetra-PEG gel, which can independently control the physical properties and can be pressed into a high-strain region without plastic deformation or fracture. Samples with almost equal stiffness and viscosity but different elasticity were prepared, with the aim of obtaining new knowledge and indices of softness perception. The experimental method consisted of rubbing or pressing a finger on the surface of the samples and having the subjects rate the "softness" and "elasticity" on a scale of 1 to 100. The experimental results showed that The integrated stress values from 0-18% strain showed a very strong negative correlation (correlation coefficient: -0.981) with the softness rating compared to elasticity.
View full abstract
-
Yuki KATAKAMI, Hiroshi HASEGAWA
Session ID: 2509
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In production, it is necessary to incorporate KANDO Quality, which is related to human emotion (KANDO: a favorable experience with Surprise that is greater than past experiences). With this, product planning is required to ensure KANDO Quality. In product planning, UQ (Uncertainty Quantification) is important to ensure quality. However, there is no UQ for emotion. This paper shows a framework for UQ and credibility evaluation in the quantification of emotional quality. Surprise (a part of KANDO) toward the same object (Stationery that has won or not won a design award) is quantified by EEG measurement, mathematical modeling, and subjective survey. The UQ in this quantification is created by kernel density estimation. The credibility evaluation was conducted using the ASME V&V40 framework, and the evidence was integrated using BN (Bayesian Network). This suggested that the credibility assessment using the ASME V&V40 framework quantitatively ensures the credibility of the UQ in the quantification of Surprise.
View full abstract
-
Ai ITO, Satoshi SHIMMORI, Yasunori HAMA, Yasushi UMEDA
Session ID: 3101
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
This study addresses the issue of knowledge and skill transfer within Japanese manufacturing companies. Introducing the concept of "Digital Triplet," the research aims to support problem-solving in digitally transformed production systems. Utilizing the process modeling language PD3 (Process Modeling Language for Digital Triplet) for explicit process description, this paper introduces a method for engineers to sequentially document knowledge and information about executed processes, suggesting an effective approach for knowledge reuse. Using a prototype system, the feasibility of this approach is demonstrated, emphasizing the importance of integrating this method into workflows for continuous technological advancement.
View full abstract
-
Shigeru HOSONO
Session ID: 3103
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Conventional web services have been designed based on an architecture in which data is stored and controlled on servers in the cloud. The services are accessed from users’ ICT devices such as PCs and smart phones that act as clients. This widely accepted architectural paradigm will be shifted by an adversative approach. Emergent technologies, self-sovereign identity and blockchain, can contribute to form a decentralized architecture that makes conventional servers unnecessary. This makes it possible to realize a self-sovereign service in which users can store their credentials locally and transmit them to service providers on their own initiative. To realize this paradigm, this article elaborates on an architecture and service design that gradually shift from centralized mechanisms of authentication and authorization to a distributed architecture of Web3.
View full abstract
-
(Case Studies of AC Compressors, Starters, and Alternators)
Shuho YAMADA, Fumiya HORIUCHI, Masato INOUE, Mitsunobu FUJITA, Akihiro ...
Session ID: 3110
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
This study investigated remanufacturing processes for compressors of automotive air conditioners, starters, and alternators, and modeled the remanufacturing process using a functional modeling approach. By investigating and observing the inputs and outputs of each remanufacturing process, we collected the data necessary for LCA and calculated the greenhouse gas (GHG) emissions produced by each process. Furthermore, by estimating the GHG emissions generated during the manufacture of new parts and subtracting the emissions of remanufactured parts, we found that the use of remanufactured parts can reduce GHG emissions by approximately 40~88%. Furthermore, as a result of identifying the processes that account for a high percentage of the remanufacturing process, it was confirmed that the manufacturing process for repair parts to replace worn parts accounts for the largest percentage. In order to increase the GHG reduction effect through the use of rebuilt parts, we confirmed that it is effective for the original manufacturer and the remanufacturer to cooperate, share information on parts whose durability should be improved, and review the design of the parts. Furthermore, it was confirmed that efficient use of equipment used in the processes related to cleaning, testing, and drying is effective for the remanufacturers to increase the GHG reduction effect.
View full abstract
-
Developing a conceptual framework
Christian CLEMM, Yusuke KISHITA, Tatsuki WATANABE
Session ID: 3111
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
The mounting environmental problems indicate that current patterns of production and consumption are not sustainable. The digital transformation has enabled sharing economy business models (SEBM) that interlink providers and consumers of goods and services to facilitate sharing as an alternative to traditional ownership modes of consumption. Whether such SEBM attain positive environmental effects depends largely on the changes in consumer behavior and product life cycles they induce. Better understanding the interlinkages between SEBM, consumer behavioral change, and product life cycles is therefore crucial for the evaluation of the environmental sustainability of SEBM. In this paper, we present our approach to developing an assessment framework to enable quantification of these factors.
View full abstract
-
(Conceptualization based on the literature review)
Yuya MITAKE, Yusuke KISHITA, Yasushi UMEDA
Session ID: 3112
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In response to resource depletion, wastes and emissions, transformational shift from a current linear economy model (production, consumption, disposal) to Circular economy (CE) is attracted attention from academia and major global industries. CE focuses on resource reuse, maintenance, upgrades, and recycling, aiming to minimize resource use and waste while fostering economic benefit of companies. Moreover, the emergence of Industry 4.0 and new developed digital technologies has gained expectation in their role as drivers for implementing CE. Indeed, manufacturing companies have started to adopt digital technologies to their business model to implement CE principles. Given the potential synergy between digitalization and CE, the concept of Smart-circular product-service systems (SCPSS) have been proposed as a new business concept framework, integrating digital technologies into PSSs to enhance circularity and economic viability. However, SCPSS is a nascent concept, lacking precise definition and principles, causing obstacles in its practical implementation for achieving CE goals. This paper thus aims to propose the SCPSS concept by briefly reviewing pertinent literature from related fields. The study addresses the relationships between SCPSS and related business models, the principles SCPSS models should adhere to, and the essential components required for SCPSS planning. Through this investigation, the paper contributes to some knowledge for understanding SCPSS and its potential in advancing CE through digital technology integration.
View full abstract
-
Yuzuki TOYAMA, Tamotsu MURAKAMI
Session ID: 3201
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Knowledge extension and recall enhancement by IT are important to support the generation of innovative product ideas. Although there have been studies to enhance recall from databases of patent information and academic papers, we proposed a method to collect highly unknown knowledge from the Internet for efficient knowledge expansion and collection of information not limited to the engineering field. We defined "minority information" as one of the conditions for low unknown information, and attempted to sort web pages in the order that they contain descriptions of minority cases by handling the language information of each web page. The results showed that there is a relationship between the low number of descriptions on the Internet and the low degree of unknowns. In the future, we will continue to study new conditions for high unknowns and aim to improve methods for collecting linguistic information on the Internet, which varies in format.
View full abstract
-
Ryunosuke YUKAWA, Junsei HIRABAYASHI, Tamotsu MURAKAMI
Session ID: 3202
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In cognitive neuroscience, it is recognized as one pattern of ideation that past knowledge is retrieved in a manner that aligns with the problem to be solved. Previous research has reported that knowledge highly relevant to the problem tends to yield relatively effective ideas but lacks novelty, while knowledge less relevant to the problem may lead to novel ideas, albeit with a lower probability of success. In contrast, this study postulates a hypothesis that knowledge described in sentences containing both highly relevant sentence elements that ensure relevance to the problem and less relevant sentence elements that introduce novelty is comprehensively effective in generating ideas. This hypothesis will be tested through experiments to validate its effectiveness.
View full abstract
-
Kyo Fujita, Shun IIJIMA, Tamotsu MURAKAMI
Session ID: 3203
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Recently, innovative ideation of what to create at upstream stage of design process is becoming more important than how to build products and services. Under such background, studies have been reported to obtain new design ideas by providing stimuli such as words to designer's thinking by using information technology. In this study, the authors consider the possibility of effectively orienting designers thinking for new ideas by presenting them with contents that does not currently exist. As a specific approach, the authors define "non-existent sentences" as sentence expressions generated by randomly assigning words to sentence elements such as subject, verb, and object, which are not found on the Internet. An experiment to present such non-existent sentences to designers to conceive new functions and user experiences was conducted. As the result, tendency to obtain more novel ideas from non-existent sentences than from existent sentences is confirmed.
View full abstract
-
Satoshi WADA, Akira OZAWA
Session ID: 3204
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
CAE (Computer Aided Engineering) tools have become indispensable for product design. However, there are problems in reducing the time needed for modelling and numerical calculations, and in making it easy for people without knowledge of CAE to use them for design. Against this background, research on surrogate models that replace CAE with AI models, has progressed in recent years. Surrogate models have the advantage that results can be obtained in a few seconds and can be integrated into larger systems such as 1D-CAE. In this paper, we report the development procedure and the result of evaluating the inference accuracy of a surrogate model that infers the vibration characteristics (natural frequency and eigenmode deformation image) of a motor bearing.
View full abstract
-
Nobuhito KATO, Keisuke SUZUKI, Yoshihisa KONDO, Katsuyuki SUZUKI, Kazu ...
Session ID: 3206
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In this study, we utilize a deep generative model called Conditional Wasserstein Generative Adversarial Networks with gradient penalty to generate the rotor geometry of an interior permanent magnet motor, based on given performance requirements. We introduce a distortion degree as a normalization term in the loss function to achieve smooth shapes for intuitive understanding and easy manufacturing. Our results show that the generated rotor shape combines multiple features from the training data. We also demonstrate that the latent variables affect various aspects of the rotor shape simultaneously. Furthermore, we determine the validity and appropriate coefficients of distortion degree for ensuring smoothness in the generated shapes. As future work, we identify the need to prevent intersections between the magnet and steel plate regions, incorporate an error term for the input-analytical value difference, and address mode collapse in the generated shape.
View full abstract
-
Kazuki HAYANO, Yoshiharu IWATA, Hidefumi WAKAMATSU
Session ID: 3207
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In generally, the creation of surrogate models presents the challenge of preparing a large amount of training data. Regarding this problem, Integration Neural Network (INN) has been proposed that can make highly accurate predictions even with a small amount of training data. Therefore, I devised an active learning method that takes into account the characteristics of INN so that we can efficiently prepare training data when creating surrogate models. Hence, I constructed an algorithm for active learning by combining an initial learning data generation method that designs experiments to capture the behavior of the entire area to be learned and a method for selecting additional learning points based on decision tree analysis. In consequence, the algorithm met the requirements of the case study and demonstrated its usefulness for efficient data preparation.
View full abstract
-
Yuki SHIMIZU
Session ID: 3208
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
We are developing a technology that supports FTA (Fault Tree Analysis) to reduce omissions of fault factors and improve efficiency. In this study, we developed a technology for expanding the causal relationships among faults using physical formula in order to generate a mutually exclusive and collectively exhaustive fault tree. We devised a logic of expansion from formulas based on the concept that there is a causal relationship between the variables on the left and right side of a formula. In addition, we devised a technique to narrow down only factors that could physically occur by using the correlation of variables on the left and right side of a formula. By using physical formulas of a fastening component as an example, we verified that mutually exclusive and collectively exhaustive fault tree was generated from causal models based on physical formulas properly.
View full abstract
-
(Proposal of a Hazard Scenarios Generation System using Petri Net)
Miho HAMAMOTO, Yusuke TSUTSUI, Akira TSUMAYA
Session ID: 3209
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Recently, unexpected product accidents are increasing due to the multi functionalized and the globalization of products. In order to solve this matter, engineers are required to predict the hazard scenarios fully and comprehensive in product designing. Objective of our study is to develop a system that generate comprehensive hazard scenarios and evaluate them. First, we analyzed accident cases, and then picked up the factors that lead to the product accident. Then, we studied a method how to explain causal relations that bring on state transition such as products contingent on factors’ status or relation by using petri net.
View full abstract
-
Ken MIYAJIMA, Takayuki YAMADA
Session ID: 3301
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In this study, we investigate the constraints for designing a piezoelectric energy harvesting device using topology optimization based on the level set method to obtain a design that takes manufacturing requirements into account. Specifically, we focus on a unimorph-type energy harvesting device consisting of a cantilevered substrate with a piezoelectric film. By defining distinct level set functions for the substrate and piezoelectric film structures, it becomes feasible to enforce constraints on the piezoelectric film structure using the substrate structure as a reference.
View full abstract
-
Takashi HAMAGUCHI, Takayuki YAMADA
Session ID: 3304
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Product design must satisfy requirements for functionality and performance under expected conditions of an actual use environment. In recent years, increasing environmental awareness has led to a demand for low environmental impact manufacturing in the manufacturing industry. Therefore, environmentally conscious design methods are needed. In the present study, we propose a topology optimization technique which reduces the environmental impact of the manufacturing process by incorporating environmental considerations into the cutting process. The optimization problem is to minimize the volume removed by the cutting process, targeting the environmental impact of the cutting process. Since topology optimization is generally applied in the framework of a continuum, the description of the region corresponding to the removal volume was modeled so that it is equivalent to a continuum. The basic verification model confirmed that it is possible to generate a shape that reduces the removal volume.
View full abstract
-
Makoto ONODERA, Isamu TAKAHASHI, Seung Hwan C. Park, Eiji SAKAMOTO
Session ID: 3305
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In the field of cast products, the occurrence of defects such as shrinkage and cracking is a critical concern directly associated with the complexity of design structure. To mitigate the risks associated with these defects, prior verification processes have been widely employed, encompassing casting simulations, experimental studies, and expert reviews. However, these traditional verification approaches are labor intensive and prone to process rework due to inadvertent oversight. In this study, ten design rules for castability were implemented in the “Design Insight CAD system”, an automated CAD geometry verification tool. This paper presents an evaluation of the system's capability to automatically detect potential defects in an engine cover part. The results showed that the system successfully identified parts to defects, such as insufficient wall thickness, corner without fillet and insufficient distances between holes. The automated inspection process was completed in 2 minutes, demonstrating the efficiency and effectiveness of the developed technology.
View full abstract
-
Yuki ITABAYASHI, Yuuki SHIMIZU, Takashi HAMAGUCHI
Session ID: 3306
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Product development requires compatibility with various KPIs such as environmental performance,product performance, cost, and reliability. To achieve streamlining design that takes multiple KPIs into consideration, We are developing multiple KPI visualization technology and tools. In addition to visualizing individual KPIs, we have developed a guide function that presents design parameters that contribute to improving KPIs that have not yet reached their target values. Is also offers, the direction in which they should be changed. A scatter diagram display function is also provided, which provides an overview of the relationship between KPIs and design parameters. This paper reports on the effectiveness of these tools in reducing the man-hours required to search for design proposals through verification of cooling fins.
View full abstract
-
Yuya UCHIYAMA, Makoto ONODERA, Yoichi ARIMA, U OH
Session ID: 3307
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
To enhance reliability of products is an important subject for industry. Therefore system models, based on model-based systems engineering(MBSE) that visualize information of a product such as structure, function, and requirements are helpful to review the product. However, it requires a lot of time to depict system models especially for complicated products. We hence develop automatic generatiton technologies for system models for the time efficiency. In this paper, we report on an automatic generation technology for a block definition diagram that shows structure of the product, from information included in a 3DCAD model. The technnology generates a block definition diagram indicating precise similar hierarchic structure to the 3DCAD model. The technology required approximately 1.5 seconds to generate the block definition diagram.
View full abstract
-
Yoichi ARIMA, Makoto ONODERA, Tatsuya HASEBE
Session ID: 3308
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In recent years MBSE (Model Based Systems Engineering) has gained popularity as a useful methodology for overviewing the whole structure, functionality and requirements of products and identifying the impact range of design changes. However, construction of system models for MBSE is time-consuming, which prohibited the usage of MBSE in actual product design scenes. Our goal is to reduce human effort in constructing MBSE models by automatically generating SysML (Systems Modelling Language) diagrams from design documents. A combination of NLP (Natural Language Processing) technology BERT (Bidirectional Encoder Representations from Transformers) and rule-based selection is adopted to extract requirements information from design documents. Open-source software PlantUML is used to generate requirement diagrams from text scripts. We applied the technology to design documents of real rolling stock products and verified that the generated requirement diagrams can reproduce over 70% of the design requirements.
View full abstract
-
Hitomi SUZUKI, Atsushi EZURA, Jiro SAKAMOTO, Takanori CHIHARA
Session ID: 3311
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Compliant mechanism is a mechanism that allows one structure to perform the function of a mechanism by providing flexibility in the appropriate position of the structure instead of a joint. Such a mechanism can also be applied as a thermal actuator that uses thermal expansion to achieve a mechanical function. In this paper, topology optimization based on the density method is used to derive the optimal shape of a gripper using thermal expansion. However, to enable efficient production, intermediate materials, known as grayscale, must be suppressed. Therefore, we focused on "Solid Isotropic Material with Penalization" and "Rational Approximation of Material Properties" for interpolation of material density, which is a design variable of the density method, and verified its validity. As a result, the interpolation method could be applied to compress the grayscale regions. Also, the holding force per unit volume could be increased by using topology optimization.
View full abstract
-
Kazushi ISODA, Nari NAKAYAMA, Kozo FURUTA, Kazuhiro IZUI, Shinji NISHI ...
Session ID: 3402
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
When designing the structure of an automobile body, weight reduction is much-needed to improve the performance regarding dynamics and fuel efficiency. As the way of designing a lightweight body ensured the stiffness, multi-material structural design is promoted. A structure with an appropriate use of different materials can achieve higher performance in comparison of single-material one. Therefore, it is important to develop the method to design an optimal structure using multiple materials which have various physical properties. On the other hand, multi-material structural design has problems about the manufacturability. One of the issues is how to estimate the influence of joint area. In manufacturing multi-material structure, different materials need joining by the means such as FSSW(Friction Stir Spot Welding). So far, the optimal design method estimating the influence of stiffness or cost by joint area are proposed. However, the method taking into account both joint stiffness and cost have not been proposed yet. In this paper, a method for designing an optimal structure considering both joint stiffness and cost based on multi-material topology optimization is proposed. Material distribution including joint area is represented by modifying MMLS(Multi-Material Level Set) using Helmholtz type equation and Heaviside projection. As a numerical example, this method is applied to cost-mass multi-objective optimization problem to show the validity of proposed method.
View full abstract
-
Kazunori FUKADA, Kozo FURUTA, Kazuhiro IZUI, Shinji NISHIWAKI
Session ID: 3403
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Acoustic properties can be controlled by placing structures such as acoustic metamaterials on the sound transmission path. However, the geometry of structures with desired acoustic properties is complex, and their design is not easy. In this study, a topology optimization method using the density method was used to design a structure with special acoustic transfer characteristics in a two-phase material model in which a coupled acoustic structure system can be considered. The structural design of an acoustic diode in an acoustic waveguide, which flows in one direction but not in the opposite direction, was realized. The effect is also observed in the frequency range of 3200-3600 Hz for structures created by binarizing the grayscale of the topology optimization calculation results.
View full abstract
-
Yuya Kozuka, Kozo FURUTA, Kazuhiro IZUI, Shinji NISHIWAKI
Session ID: 3407
Published: 2023
Released on J-STAGE: March 25, 2024
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Topology optimization has been widely used to design high functional products and additive manufacturing is an effective method for producing the optimized structures without molds. To ensure the manufacturability of these structures, it is necessary to consider the geometric size requirements that depend on the additive manufacturing process during design phase. This study presents a new topology optimization method with texture synthesis for controlling geometric sizes, such as overhang angles and length scales. Texture energy based on texture synthesis is applied as an objective function, which can generate a new structure similar to image exemplar. To consider the geometric size requirements, we translate the domain of structural members with manufacturable geometry into an image which are used as the image exemplar for texture synthesis. A multi-objective function is proposed for topology optimization to minimize the texture energy and the mean compliance. Numerical example is provided to verify the validity and utility of the proposed structural optimization method.
View full abstract