With conventional models, in order to compare the properties of push-type production systems with those of pull-type systems, in previous works it was assumed that the ordering cycle period of the push-type system was equal to that of the pull-type system. The ordering cycle period of the push-type system is, however, usually different from that of the pull-type system in actual use. For example, the ordering cycles of the push and pull-type systems may be given every day and every three hours, respectively. The results of previous works with the same ordering cycle period cannot be applied to the above practical cases. In this paper we propose a new model in order to compare the properties of the push and pull-type systems under condition that the ordering cycle period may be different. The criteria in this paper are variances in inventory and order quantity. This is because we want to investigate the essence of the behaviors of the production systems. At first, we formulate optimal push and pull-type production systems. Secondly, we formulate a demand model in order to consider the differences in the ordering cycle periods, and apply this model to both systems. Finally, we evaluate the performance through numerical analysis. We analyze the rate of the ordering cycle period coinciding with the evaluated value of the push-type system with that of the pull-type system, which is denoted by τ^*. As the result, we show that the effect of τ to the variances is very strong. If autocorrelation of the demand is strong, then τ^* is large, however, if the autocorrelation is weak, then τ^* is smaller than 2.

View full abstract

This study proposes the concept of "Hyper Logic Space (HLS)", which provides a new mathematical framework to construct a new logic algebra. The concept of HLS originates from the study of a new psychological rating method named Fuzzy-set Concurrent Rating (FCR). The FCR method requires a mathematical framework to measure the degree of contradiction in a subject's responding values for mutually exclusive items in a questionnaire. The basic form of the HLS is two dimensional ; it has a truthfulness dimension (T) and a falseness dimension (F), both of which are mutually independent of each other. Not only in classical logic but also in modern logic, neither T nor F are independent. For example, T is represented as 1-F in Zadeh's fuzzy logic. The only exception is Mukaidono and Kikuchi's Fuzzy Interval Logic Model. In their model, T and F are treated as two independent parameters in one dimensional real number truth value space. Because their model is based on one dimensional space, the negation calculus is defined by an equation where T and F are interchanged ; i.e. 〜(t, f)= (f, t). By using HLS, a new type of extended fuzzy logic-algebra can be constructed. In HLS, it is natural to define negation calculus by the equation ¬(t, f)= (1-t, 1-f). The new extended fuzzy-logic algebra satisfies Kleene's law, while Mukaidono and Kikuchi's model does not. HLS can also treat the space of more than three dimensions. Moreover, HLS can be connected to psychological experimentation through the use of the FCR method. For these reasons, it was concluded that HLS can be a useful tool for investigators of logic systems.

View full abstract

Recently the development of information equipment including computers has advanced repidly. Accompanying it, the opportunity of applying man-machine interface using cathode-ray tube (CRT) displays has increased. It is considered that in the case of work carried out by reading information displayed on a CRT, the easiness of observing the displayed information exerts a large influence on the accuracy and efficiency of the work, fatigue and other factors. This information includes patterns called icons represented by graphical user interface (GUI) as well as text data such as characters and numbers. The easiness of recognizing information displayed on CRTs has been reported regarding the color, shape, size and so on of the display. Additionally, investigations have been carried out on information processing by humans after reading displays, especially on the maintenance of short-term memory. In order to make proper judgments and actions after reading displayed information, it is necessary to make information easy to be maintained in short-term memory. Therefore, this study aims at obtaining basic knowledge on visual information processing considering the type of visual information presentation time, presentation interval and presentation direction through experiments.

View full abstract

This paper deals with the null hypothesis test for the homogeneity of plural normal distributions. The likelihood ratio test is a usual method for such testing. The asymptotic distribution of likelihood ratio test statistics is well known, so we could plan the test based on asymptotic distribution. However, when the sample size is small, the accuracy of asymptotic distribution is not assured. In this paper, we derive the cumulant generating function for the log-likelihood ratio statistics for homogeneity in plural normal distributions. Then we consider approximations of log-likelihood ratio statistics by using the cumulants. Furthermore, we confirm that the proposed approximation is more accurate by simulations.

View full abstract

In this paper, we study the design of a buffer for self-completion assembly lines. At first, we propose to set up a multi-channel buffer in each buffer storage instead of the common one-channel buffer, so each line can adjust the sequence for leveling the load at each workstation. Next, we determine the cooperative sequences for the lines. Finally, we numerically analyze the effects of the number of channels, and analyze the effect of installing a multi-channel buffer for self-completion assembly lines.

View full abstract

The purpose of this paper is to propose a method for analyzing and improving clerical work in indirectly related sections. It is generally recognized that the productivity in indirectly related sections of Japanese firms is inferior to that of directly related sections and that in western companies. However, a methodology for identifying wastes and improving the productivity of clerical work has not been sufficiently investigated in the area of method engineering. This paper proposes a method for analyzing clerical work by steps using standard Japanese language structure. These steps are consolidated into purposeful tasks so that the work can be analyzed as a hierarchical structure of tasks and work steps. The paper further presents systematic checklists for identifying wastes and listing improvement ideas at three levels, such as eliminating tasks, eliminating work steps and simplifying each work step. These results are applied to the actual case of office work at a bank and practical improvement ideas are systematically listed.

View full abstract

In the software testing phase, many development resources are consumed to detect and remove software faults introduced in the development process. So it is very important for a manager to decide how to effectively spend testing efforts during development. In this paper, testing-effort allocation policies during a module testing phase based on an optimal release problem with delivery delay under total software cost constraints are discussed. Such cost factor considerations enable us to make release decisions as to when to transfer a software system from module testing to integration testing. The underlying reliability model describing a software fault detection phenomenon is a testing-effort-dependent software reliability growth model based on a nonhomogeneous Poisson process. We also provide numerical examples of the optimal policies.

View full abstract