This paper presents a study on the simultaneous period of hand and walking motions. When a worker is moving from one workstation to another, the simultaneous motions of hands and walking occur when the worker arrives at and leaves the workstation. The characteristics of these combined movements were revealed through three experiments. When the worker left workstation, walking motion began after hand motion was finished. The time lag between the hand and walking motions was affected by the working point of the hands just before walking. When the worker arrived at a workstation, hand motion occurred simultaneous with walking motion, thus, hand motion was relational to the walking motion. However, in some cases, hand motion was not necessarily simultaneous with walking motion, that is, there was no relation to walking motion. As a result of these experiments, experimental equations are proposed for estimating the period of simultaneous motion.

View full abstract

This paper discussed and assembly scheduling problem for multi-type printed circuit boards (PCBs) in a PCB assembly center with multiple pick-insertion heads. The scheduling problem consists of a reel positioning problem (RPP) and an insertion sequencing problem (ISP) for each type of PCB. First, we formulate the overall problem into a 0-1 integer program and show that optimal reel positioning is continuous. Then, utilizing this property, we develop a composite algorithm which alternately solves the RPP by evolution strategy (ES) and the ISP for each PCB by two-optimal (2-opt) local search heuristics. We apply the algorithm to an actual industrial problem and show that the composite schedule achieves a 20% reduction in the total assembly time of the current schedule.

View full abstract

For production-inventory systems based on periodical ordering systems, it is important to control variances in order quantity and inventory. Therefore, a periodical ordering system based on optimal control theory has been proposed. The criterion of the ordering system is the weighted summation of those variances. The formula of order minimizes the criterion, and optimal order quantity can be calculated for every period. However, it is rational for the ordering system to minimize the cost function in practice. If variance in inventory is large, then a shortage or remainder of inventory would occur and inventory cost would rise. If variance in order quantity is large, then production loss and extra production costs would rise. In this paper, we formulate a periodical ordering system minimizing the cost function that depends on variances in order quantity and inventory. We show a method for searching the formula to calculate the optimal order quantity. This solution is based on optimal control theory for a linear system with colored noise.

View full abstract

The Analytic Hierarchy Process (AHP) is used for decision-making with difficult quantification. In AHP, we decide the relative importance of each criteria and alternative, and modifiers, such as "Weak" or "Strong" are used for pairwise comparison. In this paper, we propose a method for determining relative importance by using fuzzy numbers which represent the feelings of the decision maker. In the conventional AHP, a numerical value for the modifier, which increases linearly from one to nine, is used. However, in this paper, it is assumed that the numerical value of the modifier increases exponentially. Therefore, the fuzzy numbers are expressed using a logarithmic scale. The procedure in this paper is as follow. Firstly, the numerical value applied to each modifier and the sufficiently satisfactory range of each value are informed to the decision maker. Secondly, the membership functions of the fuzzy numbers are created by these values on a logarithmic scale. Finally, the membership functions are formulated into a fuzzy linear planning problem and the relative importance of each element is calculated. Moreover, we propose a method for application to group decisionmaking. The proposed method is compared with conventional methods by numerical experiment, and the efficiency of the proposed method is shown.

View full abstract

This paper considers single-machine scheduling to minimize the total weighted earliness representing the total holding cost subject to ready times and due dates constraints, and proposes an optimization algorithm based on the branch-and-bound method. Every job is available for processing only when all preceding job operations have been finished and the shop can accept delivery of materials and parts. Every job must be completed before or just on its due date and no tardy jobs are allowed. If a job is completed earlier than its due date, then the shop holds the job and incurs the holding cost for earliness. In the scheduling problem, it is not easy even to obtain a feasible schedule, and both job sequence and completion times for all jobs must be determined since machine idle times may be inserted in the optimum schedule. In the proposed algorithm, the successor nodes of the search tree are generated according to a schedule based on the Backward-WSPT rule, and the number of assigned jobs is adjusted at each node. Computational results indicate that the proposed algorithm can obtain a better sub-optimum solution if the search is closed by limits during the calculation.

View full abstract

This paper discusses the problem of designing a production system with a combination of made-to-stock (MTS) and made-to-order (MTO) products, where rush orders for MTO products cannot be regarded as being negligible. A hierarchical production planning model is formulated for a production system composed of a two-stage fabrication/assembly process. The slack capacity is set as a decision variable at a higher planning level and the investment rate of MTO products to the quantity of MTS products for the production system is set as a decision variable at a lower planning level. We analyze how the adjustment rate for production quantity will affect the unfilled-rate for MTS products and the percentage of tardy orders for MTO products as criteria for measuring manufacturing performance if the investment rate is adjusted to meet rush orders. The following points were clarified through a series of numerical experiments : (1) The higher the investment rate, the lower the percentage of tardy orders for normal MTO products ; and (2) as adjustment to the investment rate rises, the unfilled-rate for MTS products increases under a slack capacity level even if the buffer inventory for MTS products is kept at a high level. This paper presents a basis for designing a production system by setting the buffer inventory level for MTS products, the slack capacity and determining the efficiency of the adjustment rate for production quantity.

View full abstract

In this paper, for analysis of the feeling-design relationship, we propose a projection pursuit regression (PPR) algorithm combined with Quilan's ID3,which is an algorithm to efficiently generate a decision tree. In our proposed algorithm, called "projection pursuit ID3 (PP-ID3), " objects are tested by linear projections of attributes instead of the attribute value itself. PP-ID3 is able to generate simpler decision trees since nodes are represented by the linear projections of all attribute values. In order to apply the PP-ID3 for the data about "kansei (feeling and emotion)" data collected by pairwise comparisons, we further improve the algorithm to deal with pairwise comparison data similar to the way of Guttman's method. The relationship between design components of T-shirt and adjective words "loud" or "quiet" is analysed with PP-ID3.

View full abstract

This paper proposes a method to estimate highly accurate dead-times between each explanatory variable and the dependent variable in the mutiple linear regression model. In TL estimation method which has been proposed by Miyazawa et al., optimal dead-times have been estimated based on the criterion of the coefficient of determination and t-test, through varying the dead-time parameter on each explanatory variable in the model. In the improved method of this paper, the sequence of the coefficient of determination calculated in the process of the TL estimation method is regarded as a series of signals and processed by windowing. The Parzen window is selected among various window functions. A value of dead-time corresponding to the maximum of the coefficient of determination after windowing should been considered a more accurate dead-time for the explanatory variable. The estimated model is discussed in comparison with that obtained by the use of a proto-typical TL estimation method.

View full abstract

Interruptions in machining due to the sudden fracture of a cutting tool is detrimental for constructing fully automated machining systems. This paper presents a new criterion for the designing of fully automated machining systems. Two statistical models are formulated for analyzing the productivity and production costs of fully automated machining systems. Optimum cutting conditions based on tool-life distribution are obtained. Numerical results show that a fully automated unmanned system is more economical than a semi-automated manned system when the fracture rate of the cutting tool is below a certain value.

View full abstract