In Japanese Industrial Standards (JIS) there are variable a single sampling inspection plans having desired operating characteristics in both cases that the standard deviation is known and unknown. JIS Z 9003 (where standard deviation is known) provides two kinds of assurance criteria for lot quality ; one is for the mean in lot quality, the other is for the fraction defective in the lot. On the contrary, JIS Z 9004 (where standard deviation is unknown) is only specified in the case of assuring the fraction defective. From this viewpoint, Arizono et al. proposed a variable sampling plan which assures not only the mean but also the standard deviation in lot quality when the standard deviation is unknown. However, the design method is based on log-likelihood ratio statistics and is very cumbrous and circuitous. In this paper, by applying Patnaik's approximation to the sample data of the lot quality, a very simple method is proposed for designing a variable single sampling plan which assures both the mean and standard deviation in lot quality without log-likelihood ratio statistics, and the effectiveness of the proposed design method is verified through simulations.
In this paper, we propose a scheduling method dealing with transportation time by automated guided vehicles (AGVs) in a production system. AGV transportation is classified into linear type, loop type and network type. We also formulate the travel distance of an AGV, which is controlled by three traffic control patterns, ground calling type (GCT), ground calling & specifying destination type (GCSD) and from-to calling type (FTCT). Firstly, scheduling is planned by only processing time and the critical transportation between work stations is obtained from this result. Secondly, according to the facility layout previously installed, the critical transportation time is calculated. From the time when the critical transportation occurs, rescheduling is obtained using both forward and backward methods. Finally the mean flow times resulting from the forward and backward rescheduling are compared and the scheduling which has the smallest mean flow time is selected. Simulation results on its application to the sample problem are discussed with regard to scheduling performance and the effect of transportation pattern and scheduling priority rules.
Total actual flow time is known as an effective evaluation measure under Just-In-Time environments. In such case, minimization of the total actual flow time has been previously studied, but an algorithm which minimizes total processing time after minimization of total actual flow time has never proposed. This paper proposes an optimal scheduling algorithm in a parallel machine scheduling problem to minimize total processing time on the assumption that total actual flow time has been minimized beforehand under a Just-In-Time environment. The optimal algorithm is developed on the basis of an active new node search procedure utilizing the Branch and Bound method. Some computational experience shows the usefulness of the algorithm, which is coded in C-language.
Printed circuit boards (PCB) continue to become more diverse to meet various market needs. Assembly defects, solder defects and other manufacturing defects continue to appear as the result of increasing mounting densities that call for high-density, complex work in accordance with the miniaturization and lightweight requirements of corresponding products. Among such manufacturing defects, the prevention of problems related to part mounting positions and design in the design stage are required. Against this background, a computer-aided quality control system, with the goal of reducing the defect rates and improving productivity, was developed to analyze defect data by fault trend analysis through the application of a hierarchical cluster analysis. The system extracts fault trend data from the characteristics of the boards and parts. Using this trend data, instructions to prevent defects are given to the designers and the retrieval of boards and parts similar to the defective boards is prevented. In this paper the results of investigations in fault trend analysis and the processing method are described.
This paper considers a single machine scheduling with reducible processing times, and with ready and due times constraints. Every job is available for processing on the ready time when the job needs to finish all preceding operations, and cannot be processed before the ready time. Every job must be completed before or just on the due time and no tardy jobs are allowed. The processing time of each job should be determined as being within a given interval between the minimum and maximum times. The production cost for each job depends on the determined processing time and increases in proportion as the processing time decreases. Three heuristic algorithms are developed to determine the processing time, starting time and completion time for every job subject to ready and due times constraints so as to minimize approximately the sum of production cost and holding cost for earliness. Computational experiments are shown for the three algorithms.
There are many learning algorithms of layered neural networks. Recently, as one of them, Kimura et at.improved the back-propagation algorithm by considering that there is mutual correction among the weight and bias directly connected to the unit based on an extended Kalman filter. On the other hand, Rumelhart et at.attempted to introduce inertia terms into the back-propagation algorithm for the purpose of accelerating learning. In this paper, the introduction of the inertia terms into the algorithm proposed by Kimura et at.is considered and the effect of introducing the inertia terms is examined.
Suppose that a job-shop production system consist of a sales center with order-selection and a production center with switch-over. Job orders have variable estimated prices and arrive frequently at irregular or random intervals under a changeable production capacity. The macro-level problem considered is summarized as follows. Accepting all orders is not always the best option due to changeable capacity considerations. A manager of job shops is faced with the problem of selecting and processing the orders according to some decision rule, and that of increasing the idle (opportunity) cost of a job shop. We will here decide the structure of the best decision rule under a dynamic job-shop production system. This paper discusses a typical dynamic job-shop model, in the case where fixed switching costs are assumed when two processing types are switched between each other. Four control variables are introduced : two selection criteria (c_1,c_2) and two control levels (i_1,i_2). First, a stochastic model is proposed to derive the two sub-objective functions : the mean accepted price and mean operating cost of the long-run average. Next, by numerical considerations, a cooperative vs.non-cooperative problem of sales and production centers and a periodic vs.dynamic model are discussed. Finally, a revised version of dynamic job-shop production is proposed, and the non-monotonicity of selection criteria (c^*_1,c^*_2) is pointed out.