A real-time decision-making of maintenance using fuzzy agent

https://doi.org/10.1016/j.eswa.2008.01.087Get rights and content

Abstract

In many manufacturing processes, real-time information can be obtained from process control computers and other monitoring devices. However, production control problems are frequently accompanied by certain and uncertain conditions. Problems with uncertainty conditions generally include difficulty in identifying an optimal solution in real-time using conventional mathematical approaches. This study presents a fuzzy logic approach for decision-making of maintenance. Some linguistic variables and rules-of-thumb are used to form the fuzzy logic models, based on the domain experts’ experiences in production line and maintenance department. The historical production data are used to train and tune the fuzzy models. The tuned fuzzy models are then embedded into an internet-based and event-oriented information system as fuzzy agent. The production controller can easily make suitable production control decisions based on the inference results of fuzzy agents to satisfy the quick response requirement.

Introduction

Globalization and stringent competition have recently forced companies to adjust their business management methods (Ng and Ip, 1998, Petrovic et al., 1999). Many manufacturers have been driven to shift their production strategies from build-to-forecast to build-to-order or configuration-to-order to lower inventories and attain a beneficial position in a global supply chain (Chen, Lu, Yu, Tzeng, & Chang, 2003). However, satisfying the needs of quick response, the limitations of short lead times, numerous product types and small order lot sizes has led to severe control and logistics problems in production systems.

Although shop floor control is a well-researched field, and is the subject of many published articles and books, classical production control theory has been little used in real-time production environments (Stoop & Wiers, 1996). The production information in many shop floor controls can be obtained from process control computers and other monitoring devices, but most existing controlling models are unable to utilize this information to make production decisions effectively in real-time. However, this information can be easily added into the working database using a contemporary networking system. A decision support system can thus adopt the information on-line to make real-time decisions automatically.

In practice, numerous variables need to be considered in making production decisions (Kulak & Kahraman, 2005). For example, a maintenance decision is based on variables including the availability of spare parts, maintenance personnel skills and experience, human resources and maintenance strategy (Adebiyi et al., 2004, Löfsten, 2000, Swanson, 2001). However, these variables imply certainty and uncertainty conditions, particularly in a dynamic changing environment (Hotati, 2004). The most common problem with the uncertainty condition is difficulty in finding an optimal solution using the classical control theory (McCauley-Bell, 1999).

To satisfy the needs of quick response in a strong competitive market, this study presents a fuzzy agent system using real-time information to enhance production decisions. First, some linguistic variables and rules-of-thumb based on the experiences of domain experts are employed for forming the fuzzy logic model. The historical production logs are adopted to train and tune the proposed models. The modified fuzzy models are then embedded as fuzzy agents into an internet-based and event-oriented information system. The fuzzy agents can be triggered to create certainty information when data change events from the networking system occur. Finally, the production controller can utilize this information to make appropriate real-time decisions.

The rest of this study is organized as follows. Section 2 describes the problems in production control and maintenance. Section 3 then presents the framework. The working principles of the proposed model are then described in Section 4. Following, Section 5 introduces a case study to demonstrate the proposed model. Finally, conclusions are drawn in Section 6.

Section snippets

Review of the literature

High productivity and quick customer’s response are essential to most manufacturers. Quick response refers to the speed-to-market of products that move rapidly through the production and delivery cycle, from raw materials and component suppliers, to manufacturers, to retailers and finally to end consumers (Perry, Sohal, & Rumpf, 1999). Considering a competitive market requiring quick response, the production controller must quickly make appropriate decisions concerning issues such as job

The framework

To perform real-time decision-making, an agent-based approach for integrating fuzzy models is developed to assist the production controller in solving the uncertainty in production control problems. The related components are constructed on an internet-based infrastructure. Fig. 1 describes the proposed framework in detail.

The proposed framework is composed of a Fuzzy Agent Server, Database, Data Updating Monitor, Data Updating Agent, Fuzzy Executing Agent and Reporting Agent. The Fuzzy Project

The working principles

The user must first self-define the fuzzy project of the concerned problem based on his domain knowledge and/or rules-of-thumb. The historical data should be employed to test and modify the constructed models once the fuzzy prototypes are completed. This study does not provide any recommendations for the model construction. However, a reliable fuzzy project must be carefully and repeatedly tested and tuned before being considered in the decision-making stage.

Second, a verified fuzzy model must

Background on the study case

A motor engine manufacturer was adopted as a case study for implementing a test project. The chosen manufacturer has been in business since 1966, and has over 200 sets of manufacturing machines and equipment. Since many machines are worn out and ill-maintained, the production controller has one important but unhappy daily task for maintaining smooth production capability. In the event of a machine breakdown, the production controller needs to switch the current job to another machine

Discussion and conclusions

Managers today have to work toward enhancing their competitive position by providing timely responses to inquires, and delivering quality products in a timely manner. The main goal of manufacturing and distribution systems is to implement cost-effective on-line, real-time information systems. Real-time information system can assist managers to make timely and appropriate decisions in terms of daily managerial problems. However, managerial problems are often accompanied by certainty and

Acknowledgment

The authors would like to thank the National Science Council of the Republic of China, Taiwan for financially supporting this research under Contract No. NSC94-2213-E239-016.

References (39)

  • H. Löfsten

    Measuring maintenance performance – In search for a maintenance productivity index

    International Journal of Production Economics

    (2000)
  • E.H. Mamdani et al.

    An experimental in linguistic synthesis with a fuzzy logic controller

    International Journal of Man–Machine Studies

    (1975)
  • J.K.C. Ng et al.

    The strategic design and development of ERP and RTMS

    Computers and Industrial Engineering

    (1998)
  • M. Perry et al.

    Quick response supply chain alliances in the Australian textiles, clothing and footwear industry

    International Journal of Production Economics

    (1999)
  • D. Petrovic et al.

    Supply chain modeling using fuzzy sets

    International Journal of Production Economics

    (1999)
  • H. Pham et al.

    Imperfect maintenance

    European Journal of Operational Research

    (1996)
  • Y. Qian et al.

    Implementation of knowledge maintenance modules in an expert system for fault diagnosis of chemical process operation

    Expert Systems with Applications

    (2005)
  • P.A. Scarf

    On the application of mathematical models in maintenance

    European Journal of Operational Research

    (1997)
  • A. Sokołowski

    On some aspects of fuzzy logic application in machine monitoring and diagnosis

    Engineering Applications of Artificial Intelligence

    (2004)
  • Cited by (0)

    View full text