Fuzzy agents for product configuration in collaborative and distributed design process

Abstract

Design for product configuration is an inherent collaborative and distributed process. It is characterised by fuzziness of information, fuzziness of knowledge and fuzziness of interactions. Designs for configuration organisations are heterogeneous, dynamic and fuzzy evolving systems. This paper proposes a fuzzy agent-based approach to assist the product configuration. Four heterogeneous and distributed domains: (a) requirement, (b) functional, (c) solution and (d) constraint, are considered. Based on the distributed fuzzy models, fuzziness of interactions, a fuzzy computational approach for product configuration is developed. Agentification of the configuration approach, modelling and the implementation of a multiagent system, are presented. The fuzzy consensual solution agents emerge from fuzzy interactions of fuzzy distributed agents. The optimal product configuration emerges from affinities of the fuzzy consensual solution agents. A case study is presented to demonstrate the potential of this approach.

Introduction

The concept of product configuration has appeared for the first time in the artificial intelligence domain at the beginning of ‘80s in order to resolve the configuration of technical systems and was the subject of multiple researches since then [1]. Design for configuration is the process which generates a set of product configurations based on a configuration model [2], [3]. It is characterised by a configuration task.

Mittal and Frayman [4] define configuration as a set of components and a description of the connections between the components in the set. Brown [5] considers the configuration as an association and arrangement problem. The task of configuration consists of selection, association and arrangement of components and of evaluation test as well. Günter and Kühn [6] integrate in configuration task the objectives to be accomplished by the configuration process and the control knowledge in relation to the configuration process. Tiihonen et al. [7] limit the configuration to an after-sales process, where the configuration task is reduced only to the search, selection and consistency check of product configurations. Zhang et al. [8] discuss configuration-oriented product modelling and product knowledge management for made-to-order. Veron et al. [9] develop the concept of feasible configurable product in a constraint satisfaction problem approach. Aldanondo and Vareilles [10] considering product configuration as a constraint satisfaction problem, propose a global configuration approach covering the requirements, product and process configuration. Li et al. [11], Jiao et al. [12] and Liu et al. [13] use genetic algorithms to optimise product configuration.

Even so, most of conducted research considers the configurable product modelling as an arrangement problem of a predefined set of components into a valid product structure. Configuration process is seen principally as a combinatorial problem and the emphasis is put on how to decrease the complexity of this process. Most of approaches are not focused on how to design the product configuration from engineering design point of view, but rather on how to combine pre-defined components into valid product configurations. From a holistic view, there is still much to be desired in order to achieve system-wide solutions for product configuration and platform-based product development [14].

Design for configuration problem is itself collaborative and distributed. In such a process, large quantities of information and knowledge are widely distributed across multiple actors [15], [16], [17], [18]. During the design process, the amount of information available about the configurable products increases, and it becomes more complete and certain. Fuzziness, therefore, is a feature of design for configuration process. Design for configuration process is also characterised by intensive interaction between distributed actors. Designs for configuration organisations are heterogeneous, dynamic and fuzzy evolving systems.

Multiagent systems are the best way to characterise or design distributed computing and heterogeneous systems [19]. As such, the development of multiagent systems (MAS) to assist engineering design of configurable products constitutes an important step in the direction of holistic view of intelligent design. Following up this motivation, this paper proposes fuzzy agents to assist the collaborative and distributed design for configuration. The remainder of this paper is organised as follows. In Section 2, the properties of the collaborative and distributed design for configuration are analysed and a configuration approach is proposed. In Section 3, based on the mapping properties and the agentification of proposed configuration approach, a fuzzy multiagent system is modelled. The architecture of the fuzzy multiagent system platform, called APIC (Agents for Product Integrated Configuration), is developed. An application of product configuration is proposed in Section 4. Finally, in Section 5, the discussion and conclusions of this research are presented.