A. Márkus

Emergent synthesis methodologies for manufacturing

This paper offers a concise overview of new manufacturing methodologies that are based on emergent synthesis. Starting with the conceptual questions concerning analysis, synthesis and emergence, it classifies the difficulties of synthesis problems with respect to the incompleteness of human knowledge on the environment and of the specification of purpose of the artifact. Then, it clarifies the importance of emergence and self-reference and their relations to solving synthesis problems. It reviews about 300 papers according to the problem classification and in terms of three phases of concept formation, theory and application. It also discusses the system-theoretical aspects of artifactual environment.

A Market Approach to Holonic Manufacturing

Abstract The concept of holonic manufacturing is based on the cooperation of autonomous, functionally complete entities with diverse, often conflicting goals. The paper introduces a market mechanism for coordinating the activities of intelligent agents that pursue their own interest by operating under bounded rationality in a changing, hardly predictable environment. The market model is used for solving dynamic order processing and scheduling problems: conflicts between local scheduling agents are resolved by negotiating and bargaining on simple common terms of tasks, due dates and prices.

Genetic algorithms in process planning

Abstract The paper discusses the flaws of classical planning methods and outlines a new approach by which a large portion of domain-related knowledge can be represented and passed to a learning method using genetic machinery.

An agent model for incentive-based production scheduling

Abstract In agent models of manufacturing it is hard to reconcile autonomy and cooperation. The paper proposes a solution to this problem within the context of a dynamic production scheduling problem. Our organizational model is based on economic concepts, uses market rules and an incentive mechanism. The solution integrates dynamic order processing, advance least-commitment scheduling and dispatching.

Project Scheduling Approach to Production Planning

Abstract We suggest a novel approach to modelling and solving production planning (PP) problems in make-to-order production environments. This approach unifies the capacity and the material flow oriented aspects of PP. Orders are modeled as projects that compete for limited resources. The projects consist of networks of variable-intensity activities which may require several resources at the same time. The goal is to generate production plans that satisfy all the temporal and resource constraints and minimize additional or external resource usage over the planning horizon. The model is applicable at different aggregation levels of PP, as we show with two real-life case studies. Although the capacity and material flow aspects of PP are coupled, a special solver developed for the project model efficiently solves problems of real-life sizes.

Constraint-based process planning in sheet metal bending

Abstract The majority of research dealing with computer-aided process planning of sheet metal bending approaches this problem as heuristic search. Since relevant engineering knowledge mostly consists of declarations that prohibit collisions and just a handful of generative rules, there are no useful means to drive these heuristics. In order to find a new way, we have made experiments with a constraint-based approach: using predefined constraint types and geometric constraint satisfaction, complex bending problems have been solved. By returning not just a single solution but a Pareto-optimal set of solutions (i.e., operation sequences, with appropriate part orientations and tools assigned) we have left the engineers freedom to apply further, not yet modeled parts of their domain knowledge.

Product line development with customer interaction

Abstract While customizing their products, manufacturers attempt to fulfill specific requirements of the customers within the constraints dictated by the manufacturing (design, planning and production) environment, and by the economical necessity of earning profit. This paper offers a generic framework that captures more technical features of this problem than the marketing models: in addition to customer welfare and profit maximization considerations, engineering aspects are made operational, too. Driven by the interaction between customer preferences and the reallocation of manufacturing resources, viable product families emerge from a variety of technically feasible product alternatives.

Spline interpolation with genetic algorithms

A general framework is set up for the application of genetic algorithms in curve design. Then, within this scheme, the problem of spline interpolation-a frequently used method for representing complex geometrical shapes in CAD/CAM systems-is dealt with. While the method itself is simple and robust, it suffers from the drawback that some parameters must be given that are needed for the mathematical description but are not closely related to the geometrical input data of the object. The authors suggest taking a genetic approach to define the above parameters. The resulting curve minimizes a nonlinear functional which simulates the shape of an elastic bar.

Process planning with genetic algorithms on results of knowledge-based reasoning

The paper presents a new approach to solving even unusually complex problems of computer-aided process planning (CAPP). Main principles are based on the analysis of the roles and relations of CAPP in advanced manufacturing systems. Planning is carried out in several stages, by using both domain-specific knowledge and domain-independent, general methods. First, by knowledge-based reasoning, repertoires of plan elements and technological constraints are created that the final solutions must meet. Then, a global plan merging process generates plan variants so as to meet the given optimization objective. Among the various plan merging methods, one using genetic algorithms stands preeminent by its performance and robustness. Domain-specific knowledge, methods of reasoning and plan merging are described in detail.

Experiments with the integration of reasoning, optimization and generalization in process planning

Abstract The paper describes a feature-based process planning model and planning methods as verified in a prototype system with the following distinctive properties: (1) reasoning over a detailed model of the part, the manufacturing processes and resources; (2) detection of conflicts and combinatorial conflict resolution among pieces of technological advice; (3) optimization of manufacturing costs with a genetic algorithm. We show how process plans generated in this way can be generalized by omitting incidental details. Genetic optimization produces populations of close-to-optimal plans that hide some common characteristics. We discuss how such implicit similarities can be recognized and exploited so as to make the planning process more efficient in the current framework. While the focus of attention is on the fundamental assumptions, the technical feasibility of the approach is illustrated by an example.