What is the existing expert system

Knowledge-based configuration of components for automatic assembly

The assembly of multi-variant series products generally requires a large number of different, complex feed, gripping and joining movements. Automation of these movements is both complex and costly and is therefore often withheld by production companies. This results in an even lower degree of automation in assembly compared to other production areas, which can mainly be attributed to a deficit of special knowledge about automation, especially in small and medium-sized companies.

In order to properly select and configure suitable system components for an existing assembly task, a great deal of specialist knowledge is required, which in most cases is either lacking or difficult to obtain. The aim of the project is to remedy this lack of knowledge about the optimal design of an assembly system and to make this systematic and widely applicable knowledge available to as many interested parties as possible in the form of an expert system. Since the ideal technical assembly solution depends on many boundary conditions that mutually influence each other, this knowledge is difficult or cannot be algorithmized, so that a knowledge-based system can be used for this problem area. An expert system makes it possible to store the knowledge in declarative form instead of in procedural form, as in the previous program systems, which enables simplified operation combined with greater modularity (Fig. 1).

Fig. 1: Structure of an expert system

The expert system MONEX was therefore developed in the machine tools and automation technology work area, which is able to configure an individually optimized assembly system for each potential user in direct dialogue with the user (Fig. 2). In order to deal with this complex problem with an expert system, a completely new knowledge-based system was developed, which was designed according to the requirements. Relevant information, such as definitions and rules that are required to set up an assembly unit or an entire assembly system, is stored in the system's knowledge base. Based on the properties of the parts to be assembled, which the system asks the user in the form of screen forms, the expert system begins with the analysis of the assembly task. This is done on the basis of the rules that represent the specialist knowledge of the experts about automation processes and are automatically applied.

Figure 2: Concept of the "MONEX" expert system

In order to be able to take into account the specific wishes of the respective user as well as the boundary conditions of the specific automation task, there is the possibility of the weightings of the most important criteria in assembly automation, such as B. Flexibility, availability, cycle time, integration capability, ease of maintenance and noise emissions to take care of yourself. On the basis of the rules in the knowledge base and the existing generally applicable facts and definitions, with the aid of the case-specific inputs, solution requirements are created in the form of specification sheets that describe the required automatic assembly system. With the help of these requirements, the system proposes solutions in a subsequent step by comparing the requirements for the assembly system for the individual subtasks with possible components available on the market. These can be queried by the user in varying degrees of detail.

Since the data on all possible assembly components is very extensive, the expert system is linked to a database that was created under ORACLE. This database contains information about commercially available automation components such as handling devices including grippers with changing systems, transport and interlinking systems, as well as various types of peripheral components such as order and feed devices.

The database contains all the features of the individual components that are relevant for an appropriate selection, so that a well-founded evaluation of the individual alternatives including an individual ranking can be output by the expert system as the result of a consultation session.

Dr.-Ing. D. Knüpffer

Literature: D. Knüpffer

• "Knowledge-based configuration of automatic assembly systems"
• Diss. TU Hamburg-Harburg, AWA Progress Report Vol. 5, Hamburg 1992

Technical University of Hamburg-Harburg
Manufacturing Technology II
Dr.-Ing. D. Knüpffer
Denickestrasse 17
21073 Hamburg