Photovoltaic power generation has been done for a variety of sites. It is necessary to design the arrangement to consider the solar orbit dependent on latitude to increase the amount of light received by PV arrays. PV arrays can be packed into a site through arbitrary placement. However, this reduces the maintenance space for PV arrays. In addition, the shade created by front PV arrays reduces the amount of light received by behind PV arrays. Thus, packing PV arrays into a site is not the best solution. This study proposes optimization placement of PV arrays on arbitrarily shaped sites to maximize the amount of light received in arbitrary placement. The design variables of the PV array are the tilted angle, azimuth angle, and placement position of each PV array. The optimization problem to increase the total annual amount of light is solved using a genetic algorithm. The shaded area is calculated by ray tracing the sun rays. The effectiveness of the proposed method is demonstrated from several example analyses.

Guardrails are essential structures for protecting vehicles and saving human lives when traffic accidents occurred. When a vehicle collides with a guardrail in the traffic accident, some kinetic energy of the moving vehicle can be absorbed or reduced by the guardrail according to its energy absorption capacity. In the present work, we aim to investigate structural design optimization of guardrails based on collision simulation to enhance their energy absorption capacities. To reproduce a real collision condition for structural design optimization of guardrails, in the collision simulation of a vehicle collides with guardrails, we construct the finite element model of the guardrail considering details of the shape of the guardrail, the bolts connecting the guardrail and the post, and the efficient model of soil. A verified finite element model of vehicle is adopted for real collision simulation. According to the results of collision simulation, we perform structural design optimization of the guardrail using the optimization method of central composite designs, where three design variables representing its cross-section shape are chosen. For the results, collision simulations using the constructed finite element models of the guardrail and the vehicle were smoothly performed. At last, based on the optimization method of central composite designs, the optimal cross-section shape of the guardrail was determined, which could absorb the most kinetic energy of the moving vehicle.

As product systems become larger and more complex, their design space is increasing, and it is becoming more difficult to optimize them. In the past, hierarchical optimization methods have been proposed to solve this problem by decomposing the entire system into several subsystems and dividing and reducing the design space. However, they are ineffective for problems with strong dependencies among many subsystems. On the other hand, it has become clear that reinforcement learning can be used to efficiently search for optimal solutions to optimization problems such as solving small-scale games. If we try to apply reinforcement learning to large-scale optimization problems, the learning space of reinforcement learning increase, and optimization become difficult. Therefore, in this study, we used multi-agent reinforcement learning to reduce the learning space handled by reinforcement learning at a time. In this method, we propose an algorithm that reduces the learning space by allowing each agent to learn only the priority of a subsystem based on the relationship between the design variables and the evaluation values of each subsystem. The proposed method reduced the number of evaluations to less than 10% of the conventional number while maintaining the solution quality of the optimization.

Multi-stage hot forging is a typical manufacturing technology to produce mechanical components with complex shape. In the multi-stage hot forging, flash should always be minimized for material saving, and the uniform deformation is also preferable for high strength of a product. Billet shape directly affects the flash, and it is important to determine the billet shape minimizing the flash. In addition, the process parameters in multi-stage hot forging such as die stroke have an influence on the flash and the uniform deformation. In this paper, the billet shape and the process parameters in multi-stage hot forging is optimized so as to minimize the flash and the equivalent strain. Therefore, multi-objective optimization is performed. The three-dimensional numerical simulation is so intensive that sequential approximate optimization using radial basis function network is adopted for the design optimization. It is found from the numerical result that the trade-off between the flash and the equivalent strain is clarified, and the flash is drastically reduced compared to the conventional billet shape.

We aimed to develop a method to assist knowledge management of design development and maintenance in PLM in the automotive manufacturing industry by utilizing text mining. By systematizing the knowledge between parts described in the ECR and linking it with departmental knowledge, we have established a method to systematize the knowledge between parts described in the ECR and link it to departmental knowledge. Failure propagation knowledge was modeled, and a system was developed to transfer this knowledge to young engineers. By combining the knowledge of texts from different ECR and FPR situations, we proposed a relationship between parts that can affect the entire PLM.

In recent years, Japanese manufacturing industry has been facing a skill succession problem due to the retirement of skilled workers and a shortage of young workers, despite the increasing complexity of technology. To solve this problem, there is a need to extract and utilize tacit knowledge and skills, but a method for utilizing disconnected information has been expected but not yet established. For instance, in the field of product failure, a method for appropriate integration of such fragmented information is expected to obtain systematic information about failure propagation. In this study, we obtain groups of failure cases with similar failure mechanisms from past automobile failure records (FPRs) by the following steps: 1. create a network among words such as part names or failure types (failure BOM), based on a single FPR; 2. create a network among FPR documents based on the common nodes of the failure BOM; 3. apply community detection algorithms to the network among FPR documents. As a result, comparing with the manually assigned labels about symptom and cause, the resulting community boundaries matches the label boundaries. Under the assumption that the failure BOM represents failure propagation, the proposed method can be evaluated as an FPR classification based on the likelihood of failure propagation. The proposed method is effective in indicating propagation-prone failure and in getting partial failure BOM based on failure propagation likelihood, which are expected to aid human analysis of faults.

When specifications for a merchant vessel are prepared, they are based on existing specifications, and specifications are repeatedly changed in response to the shipowner's requests. Since the components of a ship are complex and interrelated, there is a possibility that some specification items will be changed in accordance with certain specification changes, but the relationships among these specification items are not organized and often depend on the intuition and experience of skilled workers. In this paper, the relationship between specification items for the equipment part of a merchant ship design document is calculated using the Cramer’s coefficient of association and expressed as a network structure so that even a non-skilled person can efficiently create specifications. The related values between specification items were derived on two scales, the components and their attributes, and multiple calculation methods were considered. It was also suggested that clustering specifications in advance and collecting specifications with similar characteristics may enable more accurate calculation of the relationship between specification items.

Our research group is currently studying and developing listening services using spoken dialogue agent and IoT technologies to assist the "mind" of the elderly at home. However, the user identification function, an essential part of the service, has not yet been realized. It is difficult to determine the identity of the person who interacts with the spoken dialogue agent. Although with the rapid development of the artificial intelligence field, various smart devices and services using deep learning have appeared in the face recognition technology, problems exist, including costs and computational resources to build and apply a recognition model. This study aims to develop a facial identification system using pre-learned model and spoken dialogue agent. Our key ideas include the automatic generation of learning data by spoken dialogue between the user and the agent and the acquisition and comparison of facial features using a pre-trained model. In this way, our face identification system can be quickly introduced and expected with only a general-purpose computer and a Web camera, without needing a conventional Internet connection and manual labeling of learning data.

Crop diseases are one of the factors behind the decrease in crop sales. Since the kind of crop diseases is large and there are diseases that damage other crops, early detection is required. Not only detecting disease strains needs much, but it is also difficult for new farmers to distinguish them; it is necessary to automate the early detection of diseased strains. This paper proposes a disease strain detection method using CAE (convolutional autoencoder) for onion strains. Since CAE learns only from strain images that are not disease, it is effective when there are few disease strain images. When determining the structure of a CAE, manual adjustment of hyperparameters such as the type and number of layers requires a lot of trial and error. Therefore, an automatic CAE structure determination method is required. In this study, we use an evolutionary algorithm that can find promising structures with a small amount of computational resources. The results of the experiment showed that the discrimination rate increased with generation. In the last generation, the maximum discrimination rate of 88.7% was confirmed.

The contents of laws and technical documents are difficult, the relationships between them are complex, and the number of laws and technical documents is increasing, making it difficult for users to find the necessary laws and technical documents. We propose a method to assist users in identifying necessary parts of legal and technical documents in a short time by using information such as the relationship between legal and technical documents. We created a network of legal and technical documents. In addition to the sentences that make up the documents and the words that make up the sentences, we added labels that group similar words together. This enables ordinary users who do not have expert knowledge of legal and technical documents to understand the structure of the documents and to grasp them systematically.

While deep learning technologies have been progressed and applied for information systems, the AI technologies become pervasive in everyday life. The technologies help human perception and decision making and aim to enhance our creative activities. With the emergent technologies, this paper proposes a concept of AI workshop system, which collaborates with practitioners in the service design phase. The AI-based workshop system will join a team as one of participants by interjecting adequate phrases and fostering empathy among them. The possibility of its use in social systems has been exemplified as a micro-tourism application of smart phones for subjective touring.

The Industrial Society is getting close to its ceiling and many issues are coming up, such as decreasing workforce, energy, etc. So, it is time now to prepare and develop a new society for the next generation. The Industrial Revolution changed our working style from working for ourselves to working for others by introducing Division of Labor. But if we realize AI, for example, consumes 10, 000 more energy than a human brain, we should look back and develop a self-sustaiing, yet self-satisfying society. AI is fine, but if we consider its large energy consumption, we immediately realize it should be applied to specail applications and we should foucs more attention on how we can self-sustain, yet self-enjoy. This paper discusses issues relating to this shift and points out that “Environment Sharing” will constitute the basis of the 22^nd century culture and Japan is expected to contribute a great deal to develop such “Service” culture. When the real “Service” is realized, we will be no more consumers. We will be true Customers

In the methodology of team collaborative design, some approaches utilize the Jungian psychological type, which supports understanding cognitive style. Psychological type is one of the individual-level variables of team effectiveness. However, type diversity causes not only positive interactions between members but also negative interactions. Therefore, we focused on the “mental attitude type,” which is relatively easy to be aware of and estimate, and analyzed the association between types when a negative interaction occurred. In the verification, the subjects were 167 people (122 men and 45 women) who took the subject named “Team Working Exercise.” Negative interactions were analyzed as follows. In Analysis 1, the subject identified the member’s type who caused a human error by the subject’s behavior. Then, Fisher’s exact test was implemented to verify whether the types between the parties were associated. In Analysis 2, the subject identified the member’s type who caused a human error on the subject. Then, the statistic test was implemented by the same procedure. Both results indicated the association with a significance level of 1% between the opposite types and showed that the occurrence contents were characteristic. Considering the subjects were in limited social groups, the results should be careful generalizing. However, those findings can help prevent and suppress negative interactions in teamwork.

The selection criteria and the selection results of a distance learning assignments on "mechanical design and drafting" considering the Corona Disaster are studied, which were examined through the actual confirmation work by the students in the author's laboratory. From the actual confirmation works, it was found that Onshape was effective for distance lectures on mechanical design and drafting, and for providing an environment for conducting collaborative assignments. However, it also became clear that there are technical issues to consider when using Onshape to perform the drafting of a classic manual winch. The actual lecture management was conducted in collaboration with the university's technical staff. The effectiveness of the collaborative lecture by the teaching staff and technical staff is described.

This paper points out that design education in Japan is too much technology-oriented and in addition, all courses are the same, because the Ministry of Education forces us to follow their principle. This principle is to pursue How, or Problem solving, and the reproducibility, quantitativeness and ojectivity are emphasided. But design is an activity to explore the new frontiers, i.e., to pursue What and Why. It is Goal-Finding. To achieve the real objective of design, design courses in Japan should be changed from education to learning. In this learning course, students are trained to find how new combinations can be produced or new market can be created with their current resources, just as past US Presidnet Theodore Roosevelt, who is famous as explores, too, told us “Do what you can, with what you have, where you are”. Thus, students are trained to explore the new frontiers. Such training is very effective in the days of rapidly, extensively and unpredictably changing world.

Sense of agency is the sense that I am the one who is causing or generating an action, and, through them, controlling events in the outside world. We previously proposed a mathematical model that predicts sense of agency in order to design the machine that gives users appropriate sense of agency. The objective of this research is to apply the mathematical model to a driving task. In the task, participants operate a handle to move the cursor on a display. The cursor represents the point of forward gaze in the operation of automobile. We applied the mathematical model to the task of operating the handle-cursor system to correct the trajectory of cursor in response to external force. The mathematical model predicts that prediction error for magnitude of the external force and observation precision of the cursor interactively affect sense of agency. The interaction between prediction error and observation precision was observed by two-way ANOVA with external force and cursor noise as factors in the experiment. It was shown that the mathematical model can be applied to systems with complex transformations between their own operations and sensory feedback.

Real-time measurement of the construction equipment operator's allocation of attentional resources is thought to be effective in preventing accidents caused by inattention while operating construction machinery. We tested the feasibility of measuring the allocation of attentional resources during tasks that simulate the operation of construction machinery of various difficulty levels, using event-related potentials (P300) during electrical stimulation, a type of tactile stimulation that does not interfere with the visual and auditory, senses required for operation. The P300 amplitude was shown to decrease as the difficulty of construction equipment operation increased. This result suggests that the P300 amplitude can estimate the operator’s attention margin capacity during construction equipment operation.

It is thought that interference of motor commands, task goals, and visual information between the left and right hands sides may contribute to the difficulty of bimanual movement. If both hands are controlled in mirror alinement at the midline, muscle control is the same in both hands. On the other hand, if both hands are controlled in parallel alignment at the midline, muscle control is not the same in both hands. The visual movements of hands in the horizontal and frontal plane are the same even in the parallel condition, although they are also not the same in the sagittal plane. It was hypothesized that such incongruency of muscle control and visual movements in the sagittal plane induces difficulty of motor control. In this study, rotational movements in the horizontal, frontal, and sagittal planes were measured. Movement direction of a right hand was designed as an experimental condition, mirror condition and non-mirror condition. As a result, in the sagittal plane, the errors and phase deviations between the left and right hands trajectories were greater in the non-mirror image condition than in the mirror condition than in the other motion planes. Therefore, it is suggested that it is more difficult to generate non-mirror image bimanual rotational movement in the sagittal plane than in the frontal and horizontal planes.

Cerebrum activation during a hand movement task and a visual stimulus task were investigated with a NIRS to address esport player’s skill. In this paper, results focused on prefrontal cortex are reported. In the hand movement task, a participant was asked to move one’s right and left wrists. Four experimental conditions were set: only right wrist, only left wrist, both wrists simultaneously with a mirror arrangement, and both wrists simultaneously with a parallel arrangement. In the visual stimulus task, a participant was asked to watch moving visual stimuli and imagine response to the stimuli without actual response motions. As the results, activation and deactivation concerning prefrontal cortex were inferred at some channels of a NIRS.

Users can participate in the metaverse meeting with or without HMDs. However, it is difficult to convey the non-verbal information such as facial expressions and eye contact unlike face-to-face communication. This makes it difficult for the speakers to recognize whether the listeners are listening or not. In this research, we propose the system’s design that detects nods and change the transparency of the avatar according to the number of nods to visualize whether listeners are listening to the conversation.

For the elderly, the act of “speaking” is important to maintain their physical function, and it is known that listening with nods and giving responses are important to encourage utterances. Also, in elderly nursing homes, conversations are often started when they look at images. So far, we proposed a listening system that promotes utterances by recognizing the user’s speech contents by voice input, and presenting images related to the words in speech on the display, and giving nodding reaction from the listener’s InterActor and images, and demonstrated the effectiveness. In this study, we have developed a listening system that guides gaze to the image presented by utterance using a robot that is familiar as a dialogue agent. In this development, we have developed two methods, one is to guide the user to gaze at the image while communicating face-to-face with the robot to the user, and the other is to guide the robot to gaze at the screen by the robot taking a position diagonally to the right of the user.

Currently, there are few opportunities to develop communication skills in clinical engineering training schools in Japan. Therefore, we developed a system to interact with an agent equipped with a rule-based chatbot, which was created with reference to the objective clinical competence examination (OSCE). In a previous study, a virtual hospital room system was developed for nursing students to interact with the rule-based chatbot agent. In this study, we added an avatar that reflects the user's embodied movements to the system. In the virtual space created, the agent playing the patient, equipped with a rule-based chatbot, responds to the user's voice, and the avatar uses a webcam to project the user's body movements in real time. This system allows for repetitive learning, which is difficult with face-to-face training. In addition, the avatar, which reflects the user's body movements, exists in the same space as the agent, allowing the user to view his or her own body movements objectively.

Biped robots are expected to be able to operate in a human living environment. It is necessary to generate autonomous walking, such as avoiding and overcoming obstacles to adapting to various environments with and maintaining stability. The walking behavior of biped robots is determined by the landing point coordinates and the time of each support period. Optimization using genetic algorithms can derive the optimal walking motion based on minimizing energy consumption and walking time with stability. The purpose of this study is to generate biped robot gait including obstacle avoidance motion. The genetic algorithm optimization used to generate a straight walking motion is extended and applied to generate a turning motion. The waist trajectory is made into a three-dimensional curve to smooth the connection between each gait and reduce energy consumption. Assuming that the walking environment with obstacles has been identified, the effectiveness of this method is demonstrated by the results of generating turning motions for several environments.

In mechanic engineering, it is important to suppress the influence of thermal expansion for machinery. In order to control thermal expansion, thermal-expansion-adjustable composites with negative thermal expansion materials are expected to be used. However, the optimum filler shape of negative thermal expansion material is not clear. In this paper, we present the optimum filler shapes derived from two methods: Eshelby’s inclusion model and topology optimization. Eshelby’s inclusion model is an accurate method to predict the properties of composites from the properties of constituent materials. Since this method can take the filler shape into account as aspect ratios of fillers, the optimum filler aspect ratio can be determined. Topology optimization can be used for more detailed analysis. The analysis was performed many times by giving a random initial shape and running the optimization calculation.

Lattice-like structures can be observed in many natural systems and industrial applications. Motivated by the need of such porous structure in the design engineering, the present work uses a single scale approach and proposes a level set-based topology optimization method for the lattice designing. The key idea is to introduce a maximum length-scale constraint, realized by a PDE-filter, into the reaction-diffusion equation based LSM. The proposed TO technique can be applied in the unstructured and/or adaptive mesh-based framework, and it can be coupled with the distributed computing technique to achieve high resolution designs. A numerical test case is presented to demonstrate the proposed methodology.

In vehicle body structural design, weight reduction is one of the most important issues for improving both dynamic performance and fuel efficiency. Therefore, structural design using multi-materials has been attracting attention in order to achieve weight reduction while maintaining the stiffness of the vehicle body. By employing multi-materials with different physical properties, superior designs can be obtained in various aspects such as stiffness, mass, and cost. However, these are trade-offs with each other. Therefore, it is important to develop a multi-material design method that can improve these aspects. On the other hand, multi-material design has some challenges that single-material design does not have. One of them is how to evaluate the joints of multi-material structures. The boundary between dissimilar materials must be joined by means of adhesion or welding. However, few optimal design methods exist that refer to joints. In this study, we propose a design method for optimal structures based on multi-material topology optimization. This takes into account not only the cost dependent on the quantity of materials, but also the cost of joining different materials. To evaluate this joining cost, level set method is used, since it requires a clear representation of the material boundary area. We applied the MM-LS (Multi-Material Level Set) method, which can represent the shapes of multi-materials. Numerical examples for the cost minimization problem and the cost-mass multi-objective optimization problem are presented.

Mount of air compressor for air-conditioner is a component that is widely used to reduce vibration and noise of compressor. Recent motors are operated with speed variable controller and are designed to perform in a wide frequency range to reduce total energy. Describes general conditions of mounting design criteria for wide frequency range operation, and suggest the technique of deriving the continuous optimal shape using a harmonic analysis-based topology optimization technology. The resulting frequency response function will be compared.

Part of the authors have proposed an infill structure generation method to obtain several suitable structural configurations, that is different from the structural optimization method that obtains a single optimal configuration. The method arranges the structural components fitted to the principal stress distributions for the grayscale optimum configuration obtained by SIMP method. This study applies the infill structure generation method to a compliant mechanism design problem. Through numerical examples, validity of the proposed method is demonstrated. Then, properties of several obtained structural configurations are compared.

Coated structures with lattice infill have attracted great interest from academia to industry. Functionally, the outer coating can protect its inner substrate and lattice infill structures exhibit great ability to improve certain structural performance. In this research, we propose a novel concurrent multiscale topology optimization method to design structures with layer-wise graded lattice infill covered by an outer coating of uniform thickness. The fundamental eigenfrequency is selected as the design objective to be maximized. At the macroscale, the velocity field-based level set (VFLS) method is used to achieve optimal macroscale structures with clear and smooth boundaries. Besides, making use of the signed distance property, the thickness of the outer coating can be well-maintained. At the microscale, the substrate area of coated structures is uniformly (or near-uniformly) divided into several layers to have different microstructural configurations and the popular solid isotropic material with penalization (SIMP) method is used to perform microscale topology optimization. Two different scales are bridged using the asymptotic homogenization method. Numerical results show that coated structures with layer-wise graded lattice infill we design own higher fundamental eigenfrequency than those with periodic and uniform lattice infill microstructures, due to the enlargement of the design freedom.

As a product moves through the fluid, it may experience fluid pressure variations due to usage condition changes, which may affect the performance of the product. Therefore, when designing the product, it is necessary to consider the robustness. In this research, we have developed a topology optimization technique incorporating fluid pressure variance into the objective function in order to be able to derive a structure that is robust to changes in usage conditions. In the verification model, it was obtained a robust optimum design that realizes smaller variance in maximum displacement against fluid pressure variations due to usage condition changes.

Rocket turbopump has low-pressure section and high-pressure section within a single system. The pressure distribution generates large axial thrust against the rotor. The rocket turbopump balance piston (BP) has the function to autonomously adjust this axial thrust. The operating mechanism of BP is as follows: (1) the rotor moves axial direction when the axial thrust changes, (2) the cross-sectional area of the orifice at impeller back shroud changes, (3) the pressure loss at the orifice changes, (4) The axial thrust is adjusted. Functional verification of the BP is performed by a turbopump test. Therefore, when design changes are implemented, it is necessary to return the upstream design stage. Rocket turbopump, which require high reliability, must be designed to restrain such return risk. In this study, 1DCAE, which performs functional design in the upstream design stage, and robust design, which reduces the effects of variation that occurs in the operational stage, were used. Robust optimization of BP was performed to find the optimal conditions that can restrain axial motion better than the current situation.

This paper proposes a concurrent multiscale optimization method of macrostructure topology and microstructure shapes for porous structures, aiming at maximizing the vibration eigenvalue. The multi-material distribution of the macrostructure and the shape of the microstructures are optimized by topology and shape optimization respectively. The homogenized properties of the porous materials are calculated using the homogenization method, and the homogenized elastic tensor and density are applied to the macrostructure. The optimum distribution of the porous material in the macrostructure is determined by a multi-material topology optimization using the GSIMP method. The KS function is introduced to solve the repeated-root problem that lies in the maximization of vibrational eigenvalues. The area-constrained multiscale optimization problem is formulated as a distributed parameter optimization problem, and the sensitivity functions are derived using the Lagrange multiplier method and the adjoint variable method. Based on the obtained sensitivity functions, the design variables are updated using the H¹ gradient method. The effectiveness of the proposed method for maximizing vibration eigenvalues is confirmed through numerical examples.

The improvement in design efficiency is highly demanded in recent product development. Set-Based Design method in Lean Product Development is known to be one of the effective methods that improve development efficiency. With this method, we can evaluate multiple design candidates and narrow them down as the development progresses. It enables us to save time, reduce costs and utilize the knowledge accumulation during the development process. With the aim of applying Set-Based Design method to the design process of air conditioner outdoor units, we have developed the practical strength evaluation tool using 1D CAE that supports developers for narrowing their solution space. This paper introduces the tool and some application of it.

Recently, it has been said that ambidexterity of exploration and exploitation is necessary for organizations to sustainably meet the needs of society and users in the rapidly changing business environment. However, no design methodology has been established to balance them and respond to uncertain future needs. Toward achieving both exploration and exploitation in product design and development, this paper proposes their definitions and the quantitative evaluation methods for their degrees based on the definitions. In these methods, words related to product functions and numerical values related to product attributes were extracted from design documents, and their differences from past documents were evaluated. Furthermore, to verify the proposed methods, they were applied to past products, for which the relationship between exploration and exploitation has been clarified. The verification showed that the evaluation results were consistent with the relationship, indicating that the proposed methods were potentially applicable to the evaluation of the degrees of exploration and exploitation.

When moving from concept design to prototype and mass production stages in product development, large capital investments are required, and careful decision-making is necessary. To assist in this decision-making, stage-gate methods and real options have been proposed, but they have yet to provide sufficient material for decision-making. Usually, patent searches are conducted to find operating principles that realize functions. However, it is thought that the subject of patent rights and information on related patents may provide materials to determine not only whether a concept being designed can be realized, but also whether it should have proceeded within our own organization. Therefore, we have proposed and verified a method to support concept evaluation in the upstream design stage and decision-making when moving to the mass production stage, using patent information.

In general, it is mechanically important to exclude areas with high loads as module boundaries because the module boundaries, which are component boundaries, are places that contribute to stiffness and strength. In conventional modularization, the integration of parts for productivity and cost reduction has been the mainstream and the viewpoint regarding the reduction of mechanical properties against module boundaries have not been sufficiently considered. Additionally, module candidates are extracted only from the geometric connecting information of parts. In this study, we propose a modular design method of frame structure that suppresses the decrease of mechanical performances by using DSM (Design Structure Matrix) analysis. Finite element meshes with a unit smaller than the existing parts are considered as a part, which enables precise extraction of module boundaries. Furthermore, this method adds to the DSM not only connecting information but also physical indicators of strain energy or von mises stress, and extracts module boundaries where DSM connecting strength and load state is low using spectral clustering based on graph theory. In the example of a simple frame structure, it was shown that module boundaries can be extracted in areas of low load and modularization is possible without reducing mechanical performances such as stiffness.

Automobile chassis parts are parts that require strength, and it has been difficult to use a rib structure for automobile undercarriage parts because a rib structure causes a chill structure, which reduces mechanical properties such as strength and toughness. In this study, a rib structure that can prevent the generation of chill structure is realized by using ultrafine ductile cast iron developed through joint research, which enables the use of the die casting method.

The sustainable development goals, adopted in 2015, include achievement of sustainable consumption and production (SCP) patterns as the 12th goal. To achieve SCP patterns, it is necessary not only to reduce environmental load caused by production and manufacturing but also to improve the sufficiency of fundamental human needs and the quality of life of consumers. The living-sphere approach aims to design products for a target living-sphere by determining the sufficiency of fundamental human needs among local residents through products using the fundamental human needs framework proposed by Max-Neef. This framework consists of fundamental human needs and satisfiers. Max-Neef argued that fundamental human needs are universal, but satisfiers fulfilling these fundamental human needs depend on region, culture, and time. In order to design products from the viewpoint of satisfying fundamental human needs, this study models the relationship between products, satisfiers and barriers, and analyzes how a product satisfies fundamental human needs. In a case study, we selected Japan as a target region and a kotatsu as an example. The result shows a possibility for an evaluation of sufficiency of fundamental human needs through the model.

In designing process, it is important to understand latent needs and reflect them in product functionality, but there is a lack of support methods to obtain latent needs from consumer needs. In this study, we proposed, applied and validated the application of a method to identify important and critical latent needs from a vast amount of needs by eliciting them from the three perspectives of potentiality (latentness), importance, and feasibility. The consumers’ responses were first translated into interpreted needs. Then, Degree of Latent Needs (DLN) was calculated by multiplying latentness, importance, and feasibility. From the results, we were able to say that DLN will be able to be used as supporting tools in discovering critical latent needs.

Recently, with globalization in industry, there is a concern about increase of defects caused by using products in unexpected ways. To solve this problem, a method to predict such defects in products in advance is required. We have been proposing a design support system that predicts possible defects in products at the conceptual design phase by modeling qualitatively expressed events and their causal relationships with petri nets. However, the proposed system presents a large number of possible defects, and the user must select them. Therefore, in this paper, we focus on the passage of time. First, we modify the event and causality model in terms of temporal context. After that, the time-consuming event and causality are formulated. Moreover, the time integral of an attribute is modeled. This makes it possible to classify whether the expected defect occurs immediately or after a long period of time. This classification will be useful for determining if the design should be modified.

In recent years, electric vehicle has been expected to spread, but due to various factors such as limitations on the vehicle side caused by technical specifications of batteries and lack of charging infrastructure, the spread of electric vehicle in society has not progressed. Therefore, this study aims to examine the possibility of electric vehicle diffusion methods that take both vehicle and infrastructure into account by focusing on battery usage. In this presentation, we report on a development of an urban scale battery usage simulator that uses battery installation and replacement methods, technical specifications such as the type and size of electric vehicle vehicles, and user behavior such as vehicle and charging behavior as variables.

Tread patterns are grooves or notches carved into shoe soles and tire surfaces. They significantly affect physical performance, such as ease of running and stopping. In addition, the complexity of their design and the fact that they form the external shape of a product also demands their visual beauty. This study proposes a multi-objective optimal design method that simultaneously considers functional properties and aesthetics. First, the dataset of aesthetics evaluation is prepared based on the semantic differential method, and it is transformed into a convolutional neural network model using a machine learning technique. It enables the quantitative evaluation of the aesthetics of respective patterns. The physical performance indices are evaluated by the finite elements methods as well. Then we solve multi-objective optimization problems using a data-driven topology design method with physical and sensory indices as objective functions. Finally, we apply the proposed method to an example problem about tire tread patterns to verify its fundamental validity and effectiveness.

For the transformer which is general industrial apparatus, electromagnetic noise and earthquake resistance are also important functions. Especially electromagnetic noise serves as an environmental load, so noise control is called for. In a transformer, at the time of operation, electromagnetism vibration is added to the iron core which is the main part, and electromagnetic noise occurs. First of all, an iron core stacks a thin electromagnetic steel sheet, and is being fixed with bolts. Therefore, when a fastening position and forces are not suitable, mechanical stiffness falls and it causes increase of electromagnetic noise, and earthquake-proof degradation. Then, it is necessary to predict correctly the oscillation characteristic of an iron core, especially natural resonance frequency, and to avoid resonance with electromagnetic vibration. In this case, the dynamic finite element method is generally used.

The finite element method generally predicts the mechanical property of apparatus in a design phase, and is used for problem solving. However, it is difficult to determine the material constants of the iron cores because these cores are composed of several thin electromagnetic steel plates laminated by bolting. The slippage between the steel plates and the bolt fastening force and position have a great influence on macroscopic material constants.

Therefore, in order to predict the oscillation characteristic of an iron core correctly, a suitable material constant is needed. This research considers the process of asking for this material constant efficiently. The oscillation characteristic is the eigenvalue and corresponding vibration mode of the iron core can be obtained by experimental modal analysis. This research shows that change of the oscillation characteristic at the time of changing bolt fastening force and bolt positions can be predicted by setting up a suitable material constant using steel sheets which imitated the iron core.

Multi-objective design exploration is a framework to aim to extract knowledge to make rational decisions in multi-objective design problems. Since a set of Pareto solutions obtained by multi-objective topology optimization contains various types implicitly, it is difficult to extract the useful design knowledge by conventional data mining methods based on the objective function values. This study focuses on a framework for multi-objective design exploration that reveals the structure of the solution space by deep clustering and logistic regression for multi-objective topology optimization. In the framework, the deep clustering technique classifies the Pareto solution set in design variables space based on structural similarity to reveal representative types, then the logistic regression technique identifies the classification in evaluation criteria space that explains the differences in the types, and then designers arrange these results into design knowledge. This paper discusses its basic validity and possibilities through an application to a simple design problem of the conceptual design of a bridge structure.

This paper describes the free description analysis on review comments evaluation using the tf-idf method, which is one of the machine learning methods. This study selected three types of comics YouTube video an evaluation target, performed machine learning by TF-IDF on the comment of the corresponding YouTube video as subjective free description data, extracted the feature values for each video, and visualized them using WordCloud tool. By conducting a questionnaire on these videos, subjective evaluations were obtained by SD method. Comparing the results obtained by machine learning with the results from the questionnaire, this study verifies the validity of the review evaluation using tf-idf method.

This study estimates emotion from the speaker's voice signal and utterance character string. First, the acoustic features are extracted using openSMILE from the utterance voice acquired through the microphone, and emotions are estimated by classifying them into emotion classes by machine learning. Then, morphological analysis is performed from the spoken voice using a voice recognition engine, emotions are classified from the spoken character string, from where emotions are estimated. At the same time, facial expressions, head and neck movements are tracked by image analysis of video image of the speaker. After synthesizing and converting the estimated emotions, it will project onto the avatar to express the speaker's emotions with the facial expressions and movements of the avatar.

We proposed an advanced mesh generation technique that reuses the proven analysis models by similar sub-part search. The proven CAD data and corresponding mesh data are managed as analysis know-how, and sub-parts similar to CAD data of analysis know-how is searched for a newly designed CAD. Then, by reusing the mesh data of the similar sub-parts, the technique for generating an analysis model with the same quality as the proven data was developed. However, in the process of arranging the mesh data of the similar sub-part, the position shift may occur depending on the pattern of the similar sub-part shape. Therefore, we developed a calculation method for the coordinate transformation matrix that minimizes the amount of misalignment regardless of the shape pattern, and confirmed that misalignment does not occur.