The production management system in Toyota Motor Corp. was introduced by Ohno and Sugimori et al., and is known as the Toyota production system (TPS), just-in-time (JIT) production system, or Kanban system. After its introduction, the system was researched from various viewpoints and implemented in various kinds of companies all over the world. The research studies and cases were reviewed and classified before 1994. However, after 1994, there has been no review of the recent works except for discussing the implementation of the TPS, JIT production system, or Kanban system. It is believed that reviewing and classifying recent works, and discussing the direction of future research works would be valuable. Therefore, this paper reviews research papers on the Kanban system, classifies them, and discusses the direction of future research work related to the Kanban system. There are numerous papers about TPS, JIT production system, or Kanban system; however, this paper concentrates attention only to the Kanban system. The Kanban system for JIT production is used to determine the quantity and timing to release orders for production or transportation processes in multi-stage production and transportation systems. The models of Kanban system researched are classified into the quantity-oriented model and timing-oriented model. Both types have been utilized for analyzing the performance of the Kanban system, as well as improving and applying the Kanban system. A decision variable that has much influence upon the performance of the Kanban system was found to be the number of Kanbans, and there has been much research done to determine the number of Kanbans in the Kanban system. Therefore, this paper reviews previous research of the Kanban system and classifies the previous literature into that of quantity-oriented models, timing-oriented models, and determining the number of Kanbans. Finally, the direction of future research work for the Kanban system are suggested on the basis of this review.

View full abstract

The objective of this study was to identify the differences in some kinetic parameters among task conditions during asymmetric load-lifting tasks. The lifting motions were filmed by two high-speed video cameras. The moments were measured by a force plate. Three independent variables were manipulated in this study: horizontal distance from a subject to the box being lifted at the start of lifting (initial distance: 30cm and 60cm), height lift (65cm and 105cm), and weight of the box (4kg and 12kg). All factors were within-subject variables. The motion characteristics of the wrist joints, elbow joints, shoulder joints, hip joints, knee joints, and ankle joints on both the right and left sides were obtained on the basis of the reconstructed data of these points by the Direct Linear Transformation (DLT) method. Based on the reconstructed 3-D data and the moments, kinetic parameters such as the power of both forearms and both upper arms, and spinal loads such as the compression force F_c, anterior/posterior shear force F_<sAP> and the medial/lateral shear force F_<sML> were calculated. The following characteristics of the asymmetric lifting task were revealed regardless of the task conditions: (a) the loading on the lumber spine was the largest at 30% of the normalized time; and (b) increasing the waist-twisting angle significantly increased the medial/lateral shear loading on the lumber spine. Moreover, it was clarified that the kinetic parameters were greatly affected by the three task conditions (initial distance, height, and weight): (a) the initial distance affected the F_c from 0% of the normalized time to 70%; (b) the height affected the spinal loads after 40% of the normalized time; and (c) the weight affected the F_c and F_<sAP> throughout the lifting task and affected F_<sML> after 40% of the normalized time.

View full abstract

The N-version programming (NVP) approach is a kind of fault tolerant techniques and is a software system employing functionally equivalent, yet independently developed software components. It relies on the application of design diversity (i.e., program versions are independently designed to meet the same system requirements). A consistent set of inputs is supplied to all versions and all N versions are executed in parallel. A decision mechanism must gather the available results from N versions and determine the result to be delivered to the user. When constructing a large-size software system with extremely high reliability, we must use components with much higher reliability, however, this is very expensive. Therefore a large, highly reliable software system should be designed while minimizing cost. Genetic algorithms are known as one of the most powerful optimizers or near-optimizers of combinatorial optimization problems. We propose two genetic algorithms, one employing "random key representation", and the other employing "binary representation", for solving the N-version program design problem formulated as a 0-1 non-linear integer programming problem, respectively. Further, the efficiency of the proposed GAs is proven through some numerical experiments. Then, the characteristics of each representation scheme are investigated and evaluated.

View full abstract

In the SCM age, the demand-to-supply management supplies to meet demand has become more important. However, marketing and manufacturing departments do not collaborate well with each other (Matsui and Fujikawa, 2005). Aggregate production plans are generally decided on the basis of demand forecasts, which are usually independent of production planning. Collaboration between demand forecasting and production planning is required to maximize profits. For this collaboration, a strategic demand-and-supply map is developed using the theory of pair-matrix table (Matsui, 2002). This map consists of a row for demand speed (input controller) and a column for the smoothing factor (output controller), and the respective elements show the objective indices in economics and reliability. Using this, an ellipse-cross theory was formed, having four extreme poles (strategic standpoints). This paper presents an effective planning tool to support the demand-to-supply manager, by introducing a strategic demand-and-supply map. This planner consists of demand forecasting, aggregate planning, strategic map, APS Scheduler and progressive control steps. The respective steps give the underlying theory and numerical examples using demand data. This manager-aided planner is useful in the systematic training of sales/production managers.

View full abstract

Ways for evaluating work burden in an assembly process can be classified into two methods: one for obtaining a torque that is exerted on the joints of the human body, and the other using rapid upper limb assessment (RULA), which is set in accordance with the bend angles of joints. The drawback of the method for obtaining the torque of joints is that it necessitates measuring the elasticity coefficients of a number of muscles in the limbs of a worker. On the other hand, the RULA method is not applied until after the design and construction of a work process is actually completed. This paper proposes a method that makes it possible to analyze work burden even in the process design stage. Our method employs a human body link model. This model has seven joints in the upper half of the body and two joints in the two legs. As constraints, we designate the positions of the fingers at the starting and ending points of a motion, physical characteristics and the range of bend angles of joints established by RULA. Further, we set the sum total of the values obtained by multiplying the bend angles by a coefficient as the objective function. Next, we obtain a solution that minimizes the objective function using the quasi-Newton method to reveal the postures at the starting and ending points of a motion. As an application example, we designed a working area for the assembly of a household appliance. Then we actually manufactured a work table for assembly and carried out the assembly to study the differences between the results of the simulation by this method and actual work burden.

View full abstract

This paper proposes a case-based reasoning system to evaluate customers' credit in a small-scale student dress wholesale company. This system aims at assessing a customer's credit through a five-grade credit score: score of 1 meaning a normal customer for which all orders are accepted, and score of 5 meaning an abnormal customer for which all orders are rejected. Because most of the customers are minor small businesses that do not disclose financial information, it is almost impossible to obtain their financial data. The proposed system extracts the features of customers from the database of the management information system and assesses a customer's credit on the analogy of features among all customers. The features of a customer include annual purchases, average amount of late payment, average number of days payment is late, the number of late payments, the number of months purchase occurs, etc. This information is very easy to obtain from daily transaction data such as sales, payments by customers and bills sent to customers. We provide the system architecture and a description of past cases, and furthermore, we develop a procedure to retrieve similar cases and to assign the given customer a credit score based on the credit scores of similar cases. The performance and effectiveness of the system was confirmed applying the real transaction data of the company. The credit scores for normal customers provided by the system are more than 98% in agreement with the judgments of financial operators of the company. By comparing the credit scores of the customers during 2001-2003 obtained from the proposed system and that from a linear discriminant analysis method, it is shown that the system can correctly discriminate normal customers from abnormal ones and has a higher discriminating ability than the linear discriminant analysis method. Since the default probability of minor small businesses is very high compared with major companies, we consider that the proposed system is more valuable and can be applied more widely than other existing credit scoring systems. Finally, it is very easy to coordinate the proposed system with existing information systems.

View full abstract

A large number of farming households in Japan are operating cut rose harvesting businesses on a year-round cutting schedule by using hydroponic plant cultures. There are two different varieties of cut roses: a standard type and a spray type. Both of these types can be purchased from nursery stock and then fix-planted for the production of cut roses. The yield of cut roses varies according to the season. The following are assumed: the area of the greenhouse is fixed; the period from the time of planting the nursery stock to the time of replanting is four years; and the cut roses (i.e. year-round harvesting type) are produced by hydroponic plant culturing. Then, a plan for the optimum utilization of the greenhouse based on mathematical programming that includes restrictions such as yield per month and other factors for the purpose of obtaining maximum profit is proposed for a case in which the ratios of the yields in different seasons are based on the optimum utilization of the greenhouse for the production of cut roses.

View full abstract

This paper proposes an "Evaluation and Selection Method for Research and Development Projects" in the field of research and development (R & D) management. The concept employs a technique in which multiple decision parameters are utilized to access and choose projects; especially the problem structure parameter from the viewpoint of practical application in actual situations. We first discuss the corporate executives and R & D managers that assess new projects and evaluate various ongoing programs. Project variety is usually labeled as follows: basic research, applied research, development research or improvement research. A number of assumptions are presumed when evaluating different projects, such as: 1) To what extent is the project connected to the business or technological strategy of the company? 2) How will the project contribute to current business and the prospect of future corporate growth? And, 3) Can the company complete the project utilizing its current labor force and technologies? All organizations must consider these points in order to select the best project combinations for expanding their business opportunities. Following these considerations, the company then makes a decision on R & D investments. In this paper, we propose a new method that can be used as a tool to ensure the validity of project assessment and selection. Tangible examples with numerical data are provided. It is believed that this method can greatly support the R & D investment decision-making processes of corporate executives and R & D managers.

View full abstract

A time series study conducted as part of this research confirms psychological aspects related with teamwork on group decision-making. The survey was based on Group Dynamics and focused on Group Process conducted by team members within a team using management simulation. Differences between teamwork awareness were tested statistically according to time series, at the starting point, mid-point and closing point, respectively. A management simulation survey was conducted in 1998 and 80 students answered. In conclusion, the following hypotheses were accepted. Hypothesis 1: Mean of awareness to teamwork has a larger value at the closing point than the starting point (comparison of mean value). Hypothesis 2: Closeness of awareness to teamwork changes according to the time series, and it has stronger unity at the closing point than the starting point (comparison of area and distance depending on the factor loading value position of principal component analysis.) Hypothesis 3: Stratified structure of awareness to teamwork changes according to the time series, and it has a clearer structure at the closing point than the starting point (comparison of cluster analysis dendrogram.)

View full abstract