Multi-robot systems have been introduced actively in heavy industries in order to increase the productivity. However, since robots are disposed as densely as possible in a working stage in the multi-robot system, collisions between adjacent robots could occur. Because the collision prevents effective production, collision occurrence should be avoided. On the other hand, robustness of a schedule in a sense that small time lags between the schedule and the actual operation do not affect occurrence of collisions is also important. This paper deals with the scheduling system based on the genetic algorithm (GA) which attains these two objectives. Its effectiveness is shown by numerical simulations with practical workpieces.
In real manufacturing system, changes take place in the shop floor frequently. When a manufacturing simulation system is developed and used for the daily operational planning and evaluation, the simulation model must be in a good correspondence with the real manufacturing system. Therefore, how to reflect various changes in the manufacturing system to the simulation model and thus keep the model updated become important issues to be answered. In this study, towards automated updating, simulation-modelling formalism for manufacturing system is proposed. This formalism makes use of change suppression technique as well as late binding and polymorphism in object-oriented technology to make the simulation model uniform, consistent and keep unchanging even for various changes so as to achieve the automated updating. A generic framework capable to deal with the automated updating for manufacturing system simulation is constructed. The proposed modelling formalism is proved to be effective through experimental simulation system construction.
An appropriate performance measurement system is the key to effective supply chain management. Two hurdles are present in measuring the performance of supply chain as a whole. One is the existence of multiple measures that characterize the performance of individual supply chain members. The other is the existence of conflicts between supply chain members with respect to specific measures. As a result, the efficiency of a supply chain cannot be characterized directly by the performance of supply chain members. The current study develops an approach for characterizing and measuring supply chain efficiency and achieving best practice. Models are provided to define the supply chain efficiency and to measure the performance of a supply chain as well as supply chain members. It is shown that a supply chain as a whole has potential to achieve more cost savings and a better performance through coordination and information sharing based upon our linear programming problems.
Swing-up and stabilizing control of an inverted pendulum is one of the most common experiments used for illustrating nonlinear control techniques. Since an actual inverted pendulum has limited pivotal travel, the controllers have to be designed to satisfy this constraint. However, it is difficult to design controllers for underactuated systems with state constraints, like inverted pendulums. Consequently, there have been very few studies into this problem. In this paper, we propose a control law which can swing up and balance a translational inverted pendulum through limited pivot travel. Bearing in mind that the energy of the pendulum can be controlled according to the sign condition of pivot acceleration, we develop a method for swinging up the pendulum which involves controlling acceleration of the pivot as well as limiting its travel. The proposed balancing control law is a linear one designed by applying block control methods to the linearized model about an unstable equilibrium point, which can stabilize the whole system as keeping the amplitude of the pivot small. The results of our simulations and experiments demonstrate the effectiveness of the proposed control law.
Genetic algorithm (GA) has been widely used in combinatorial optimization problems, but direct application of GA to grouping problems often turns to fail. This difficulty arises from the large amount of constraints that are intrinsic to the problems of this category. Following this observation, this paper adopts an alternative encoding based on the “grouping genetic algorithm” (GGA) proposed by Falkenauer. The whole algorithm consists of two stages : session construction and session allocation. Keywords and category keywords are used for grouping. The experiments using real data will validate the usefulness of the proposed method.
This paper proposes a rate-based control method for congestion control. Our proposed method is based on a PID controller predicting a queue length in a network switch. To stabilize the queue length for time-varying the propagation delays, we design a robust PID controller.
In this paper, we propose the training method for three layer neural network classifiers in which we calculate the weights by maximizing margins in each layer. According to the CARVE algorithm, we need to find a hidden layer hyperplane that separates a set of data of one class from the remaining training data. Then, the separated set of data is removed from the training data. We repeat this procedure until the remaining data belong to one class. In the proposed method we use a heuristic algorithm to train the support vector machine so that the data on one side of the hidden layer hyperplane belong to one class. For the output layer, we use a quadratic optimization technique. The performance of this new algorithm is evaluated using some benchmark data sets.