Being a fashionable word literally, the term flexibility has been recognized to mean a key aspect of a firm's response to competitive markets. The purpose of this paper is, through available literatures relative to mainly manufacturing/engineering, (1) to examine various flexibility concepts from diverse disciplines, (2) to re-classify the types of flexibility in view of various perspectives (3) to facilitate the understanding of the interactions between environmental uncertainty and flexibility needs, and (4) to provide some conceptual directions for intergrating them into the planning, design and implementation of the production system.

View full abstract

The data in the fields of engineering, biological, and social sciences, are often represented by cross-classified data. Such data have been analyzed by many researchers using various methods. The association models proposed by Goodman have especially become major tool. In this article, another look at the association models by a mixed additive and multiplicative model is discussed for the analysis of multiway cross-classified data.

View full abstract

Given a network and a budget, the budget network design problem consists in selecting a subset of links that minimize the sum of multicommodity flow costs subject to the budget constraint. We propose a lower bound algorithm using a Lagrangean dual ascent method and approximate algorithms using information of Lagrangean relaxation solutions. The effectiveness of our algorithms is demonstrated in computational test problems with up to 75 nodes, 2775 links and 2775 commodities.

View full abstract

We propose a method for designing multi-item assembly lines that are composed of assembly robots in the light of the installation cost. We design a line where one assembly robot fastens one part since the assembly robot has difficulty fastening different parts. First we design two types of assembly lines : a mixed model line that can assemble a several kind of products in one line and a single line that assembles only one kind of product. Next, we estimate the number of assembly lines and their installation cost needed as the amount of the production increase. Finally, we choose the assembly line that can be installed at the lowest cost.

View full abstract

In a Bayesian test, when we receive data {x}, we examine the Bayesian test by using the probability density function f(x|θ). However, these data may be fuzzy in actual circumstances. In this paper, we define these data as fuzzy interval data with human vagueness and propose a new method which can treat the fuzzy interval data by usual statistical methods. That is, we treat the central points of fuzzy interval data as representative data, and the validity of the correction proposed by us is investigated by the computer simulations under realistic situations.

View full abstract

In this paper, we study the behaviour of machining-assembly automatic transfer lines. The line consists of an L-stage series assembly line and L_i machining stages which supply necessary parts to assembly stage i. At first, a new approximation method is proposed to estimate a line production rate, and the approximation error is presented. This method has two steps. Step 1 is that an L-stage series assembly line is decomposed into 2-stage merge lines, Step 2 is that a 2-stage merge line is decomposed into 2-stage series lines. Next, using the approximation method, we numerically analyze the effects of numbers of assembly and machining stages on the production rate, and analyze the effect of installing buffer storages.

View full abstract

Recently, many companies are strongly obliged to reduce the inventory level, and these companies must examine whether they can achieve a goal of imventory level and/or a probability of out-of-stock. This paper proposes a simulation method for inventory control which includes scheduling technique in order to respect the restriction on capacity of production. This method is applied to an actual problem to investigate effects of the method.

View full abstract

A new model and its solving procedure for the commodity distribution system consisting of inventory facilities such as distribution centers or warehouses and consumer points are discussed. In this paper, demands of the consumer points are assumed to be random variables which have known probability distribution, and an integrated model is built, where both of fixed reorder quantity policy model which is one of the typical inventory policies and transportation model are considered simultaneously. The problem is formulated as an optimization problem to minimize the total of inventory related costs and transportation costs. That is, it is formulated as a 0-1 mixed integer programming problem including nonlinear function, and an algorithm based on the branch and bound method is developed, where the lower bound can be calculated by decomposition of the problem. Meaning and validity of the proposed model are made clear by a numerical example.

View full abstract

The objective of the study is to construct a learning algorithm with high speed and accuracy for a neural network model. A learning algorithm from Conjugate Gradient Method is an effective method for high speed learning. However, the Conjugate Gradient Method is unstable in its convergence depending on an initial setting of network parameter values. We, therefore, propose a learning algorithm with higher accuracy based on Conjugate Gradient Method. The method proposed in the study is based on improved formula to create a conjugate direction. We call the proposed method "Improved Conjugate Gradient Method".

View full abstract

In our former paper, we discussed the outline about how to develop a GP system which should be easy to use even by PC-users. The interest of this paper is centred around showing the detailed algorithm for solving GP models throush discussing mainly the utilization of combination of the Product Form of the Inverse with the Lexicographic Rule of Cycling Prevention. The paper also shows an attempt to expand the system for improving its practicability by combining a decomposition approach with the two-step-modelling method proposed in our paper. Finally we show an application case to clarify the applicability of the system.

View full abstract

The locational decisions of manufacturers are basically made by considering many location factors and their preferences toward these factors. In this paper, we present a fuzzy linear programming model for identification of these preference weights which can reflect the intentions of decision makers. This model was applied to the state selection of Japanese manufacturers located in the US and the validity of this model was evaluated. The use of this model has made it possible to assign the preference weights and also to select the optimal locational sites for two plants with mutual relations.

View full abstract