Traffic jams are getting more attention due to their environmental impacts in addition to resultant economic losses recently. The most popular method of traffic signal control is to select an appropriate signal pattern for the current traffic flow out of those pre-calculated. This selection process depends on experiences of traffic engineers and. social experiments currently. Furthermore, a signal control to improve a traffic flow at a specific cross section does not necessarily improve the flows of the whole area; i.e. it may cause a traffic jam in another area. To resolve these issues, we study a traffic signal control using optimization methods. We calculate the parameters of every single signal inclusively and optimize a flow as a whole. The goal of this work is to find optimal sets of control parameters of traffic signals to improve the traffic flows globally. We developed an optimization system using the genetic algorithm in conjunction with our multi-agent traffic simulator. We applied the system on traffic flows in an area in Okayama prefecture. The results shows that our approach is reasonable, but needs improvements in accuracy.

View full abstract

Miniaturization of electronic equipment requires high density packaging technology. Therefore, the concern about the reliability in electronic devices rise further more. In this study, authors research better material properties of the package board by using multi objective GA algorithm and FEM method, and analyze the results by data mining technology. Data mining results show better material properties are low thermal expansion coefficient in high temperature area.

View full abstract

A method for system identification with Linear Matrix Inequalities (LMIs) is proposed in this report. We show that system identification problems for various types of systems, e.g., an identification problem of LTI discrete time systems, continuous time systems and time varying systems etc., can be formulated as an LMI optimization problem. With the present problem formulation we can easily incorporate inequality constraints on identified parameters, e.g., a constraint on the upper and the lower bounds and/or quadratic constraints, etc.. This property helps us to get the mathematical model that is physically valid, such as the damping and the stiffness matrix of mechanical systems generally take positive value. Furthermore, the proposed LMI formulation for system identification can be applied to a framework of iterative system identification and controller design. The effectiveness the present identification methodology is experimentally examined with a system identification for a 3-story benchmark building of NCREE, Taiwan.

View full abstract

The plant seeds are moved in a variety of ways, for example, carrying on the window, insect or animal. Dispersion by bursting fruit is a typical moving way of plant seed that is self-active and independent of other creature . Fruit, which can burst powerfully and spread the seeds widely, are advantageous to expand breeding grounds of the plant in natural selection process. So, pericarp bursting mechanism of extant plants seems to be optimized mechanically and structurally. Bursting motion of pericarp depends on its mechanical stress in self-equivalent condition just before bursting. In this study, pre-burst stress distribution of fruit was evaluated by finite-element method and optimization technique.

View full abstract

Aluminum beverage bottles with a screw top of 28-mm and 38-mm in diameter have been launched on to the Japanese market recent years to meet new drinking custom trend of the consumers. However, some Japanese customers feel that the 28-mm opening is too small and also feel that the 38-mm opening is too large. Therefore, we need to develop a method to evaluate human feelings when drinking directly from the bottle opening and to identify the opening size that is best for consumers' drinking ease. Hence; we propose an evaluation function of drinking ease in order to identify the rigorous dimension of opening diameter. We defined the evaluation function of drinking ease in terms of difference between the actual flow amount from the bottles and the appropriate flow amount. If the difference decreases, the consumers possibly feel comfortable. We considered three variables (the bottle opening size, the remaining beverage amount in the bottles, and the height of subjects) and constructed the evaluation function of drinking ease using Radial Basis Function Network. Results from the optimization of the proposed function, the subjects may get the closest beverage amount to the appropriate amount and feel comfortable when they drink from the 34.3-mm opening. We have identified the rigorous dimension of opening diameter that the subjects may feel comfortable using the proposed function.

View full abstract

Concurrent engineering has presented new possibilities for successful product development. 3D-CAD systems have helped propel the concurrent engineering practice. Designers should, in general, deal with a huge variety of design knowledge and create design ideas by their intuition or experience through a trial and error process at the early phase of design. This paper proposes a design system that supports designers to create design ideas by optimizing the relationship between 3D-CAD solid models and design knowledge. Our proposed system consists of 3D-CAD system and an agent system. 3D-CAD models are added to non-geometric information that is attribute information, and we agentify the 3D-CAD solid models and design knowledge. This system is a multi-agent system which can refine the 3D-CAD models by collecting and managing design information autonomously. This paper discusses the effectiveness of our proposed system through an application to design of a Continuously Variable Transmission.

View full abstract

Recently, we have concerned the strategic optimization on logistic network design and developed an efficient two-level solution method. To cope with extremely large-scale problems, in this paper, we propose an extended algorithm for parallel computing for the two-level algorithms that utilizes the analogy between the algorithm and the master-worker configuration of PC cluster. To enhance the efficiency, we adopted a population-based algorithm like particle swarm optimization (PSO) and developed a new discrete algorithm of PSO to deal with the binary decision variables standing for open or close of the DC sites. Then, we have developed a parallel procedure that can make the over head for the parallel computing very small, and finally bring about high performance for the parallel computing suitable for the present goals and circumstances. Through numerical experiments, we confirmed the effectiveness of parallelization from a few aspects that characterize the proposed method.

View full abstract