Itsuo Hatono

Reinforcement Learning Approaches to Biological Manufacturing Systems

Abstract The concept of Biological Manufacturing System (BMS) aims at dealing with unpredictable changes in the external and internal environments during whole product life cycle, based on biologically-inspired ideas such as self-organization, learning and evolution. We previously developed evolution-based and self-organization models of the floor level that are able to cope with environmental changes such as system reconfiguration, machine breakdown and unforeseen production requests. This paper describes reinforcement learning approaches to the modeling of BMS, in order to deal with the various kinds of complex global objectives. The effectiveness of the proposed model is also discussed by simulation with global objectives.

Line-Less Production System Using Self-Organization : A Case Study for BMS

Abstract The paper proposes a novel approach to the design of line-less production systems. The approach is based on the concept of biological manufacturing systems (BMS) that can deal with complexity in manufacturing using the ideas of self-organization, evolution and learning. In the line-less production systems, all production elements can move freely on the production floor using self-organization in order to adapt to fluctuations such as the diversity of production demands and the malfunction of machines. The effectiveness of the line-less production system is discussed and its feasibility is demonstrated by computer simulation. Furthermore, a mini-factory consisting of small autonomous robots is developed as a prototype of the line-less production systems.

Emergent Synthesis Approaches to Control and Planning in Make to Order Manufacturing Environments

Abstract The main goal of manufacturing systems with make to order environments is to provide products for unique requirements of customers. The demand for products can vary not only in product variety but in optimum criteria. Whilst a manufacturing system might be capable of producing the required product, there can appear difficulties for the system to fulfill the cost and time related constraints of the consumer. This paper introduces emergent synthesis approaches to manufacturing control and production planning in make to order environments to evaluate and control the range of time and cost constraints that the system is able to complete.

Facility Layout Planning Using Self-Organization Method

Abstract This paper proposes a novel approach to facility layout planning based on the concept of biological manufacturing systems that deal with the inherent complexity of manufacturing using ideas of self-organization, evolution and learning. A self-organization method is proposed to generate facility layout plans autonomously according to the material flow which emerges from the local interactions among production elements. A case study of facility layout planning is presented for semiconductor manufacturing, in which it is difficult to find a proper layout because of complex process flow. The effectiveness of the proposed method is discussed in terms of the accumulated traveling distance of products and the throughput.

Scheduling for on-time completion in job shops using feasibility function

This article focuses on job shop type manufacturing systems where a high variety of product of different volumes are requested to be produce on a tight due date. The feasibility function (FF) is addressed in this article to schedule jobs in multi-machine random job shop, where the purpose is to minimize unit penalty by achieving a balance between the number of tardy and early jobs, and reducing the difference between the maximum and the minimum lateness of jobs. A job shop simulation model based on multi-agent architecture developed by the authors provides an environment for comparing the FF to commonly used dispatching rules. The results show the benefit of using the FF. Discussions reveal that this concept is more reliable in case of due dates with different tightness level.

Reinforcement learning approach to self-organization in a biological manufacturing system framework:

Abstract Biological manufacturing systems (BMS) aim at dealing with complexity and uncertainty in todays manufacturing, employing biologically inspired ideas such as self-organization, learning and evolution. This study proposes a self-organizing manufacturing system where manufacturing process progresses as a result of local interaction among manufacturing elements using potential fields as an implementation of the BMS idea. The study then verifies that systems feasibility. However, it is difficult to achieve the complex global objective of a system using only the self-organization method because it uses only local information of each element to implement; achievement of global objectives often requires global system information. This study proposes a reinforcement learning approach to a self-organizing manufacturing system to achieve the global manufacturing system objectives. The proposed method is applied to the maximizing throughput problem with consideration of the machine set-up time. The results of computer simulations illustrate effectiveness of the proposed method.

Multiobjective combinatorial optimization for performance evaluation by a metaheuristic satisficing tradeoff method

In this paper an effective meta-heuristic approach is proposed to realize a satisficing tradeoff method for solving multiobjective combinatorial optimization problems of performance evaluation. Firstly, Pareto optimal solutions (individuals) are generated by using a genetic algorithm with family elitist concept for a multiobjective combinatorial optimization problem. Then, we try to find a preferred solution of the decision maker based on the satisficing tradeoff method. Usually, a conventional satisficing tradeoff method needs to solve a complex min-max problem in each iteration of the algorithm for a given aspiration level of each objective function. The min-max problem is to minimize maximum value of a regularized regret function. In this paper a new meta-heuristic satisficing tradeoff method is proposed in which we do not need to solve a complex min-max problem in each iteration, but we try to find a min-max solution in the Pareto optimal solutions (individuals) generated by the genetic algorithm. We further revise the min-max solution by using a local search approach such as a simulated annealing method. As a numerical example a flowshop scheduling problem is included to verify the effectiveness of the method proposed in this paper.

Database as a Tool for Visualization of Institutional Information: A Case Study of KOBE University Information Database (KUID)

Based on a case study at KOBE University, the process of constructing an institutional information database is described. The University actually had all the relevant data, but they it was not centrally accessible as an aid to decision-making. The first step in constructing the database was to identify and review existing data stored in disparate information systems, followed by steps to identifying the principal users for the database and the principle unit of analysis to be use. The experience suggested the following design principles: common information is necessary for consensual decision-making, especially in the evaluation and assessment work, and the requirement of expertise, not only with technical skills (e.g. Information and Communication Technology), but also with a knowledge of work flow and institutional contexts is needed, along with a cooperation and coordination skills of other departments.

A decision support system for designing business strategies - an application to soft drink business

This paper deals with a decision support system of planning business strategies for demand chains. To describe the complex behavior of each element in the demand chain, an agent based modeling approach and methodology of virtual market are introduced to develop the simulation model. The proposed model is applied to softdrink vending machine business as an example. Demand characteristics of soft drinks through vending machines are investigated based on the methodology by simulation studies.

Floor layout planning method based on self-organization

Floor layout to realize short accumulated distance of product and high throughput is required. Currently, however, it is very difficult and takes a very long time to optimize floor planning in a wafer process, because of complex process flow. In this paper we propose a new method of floor planning based on self-organization to solve these problems. We verify the validity of applying this method to semiconductor manufacturing. Self-organization method can generate a floor layout plan autonomously. In particular, potential field modeling method can describe simulation models in a simple way, because it controls all entities in the same method. The results of simulation indicate that the proposed method can provide the layout plan with short accumulated distance of product without requiring considerable labor and time.