Data processing: generic control systems or specific application – Specific application – apparatus or process – Product assembly or manufacturing
Reexamination Certificate
2001-05-16
2003-12-30
Picard, Leo (Department: 2125)
Data processing: generic control systems or specific application
Specific application, apparatus or process
Product assembly or manufacturing
C700S104000
Reexamination Certificate
active
06671568
ABSTRACT:
BACKGROUND AND SUMMARY OF THE INVENTION
This application claims the priority of German Application No. 100 23 668.5, filed May 16, 2000, the disclosure of which is expressly incorporated by reference herein.
The invention relates to a method for supporting a multistage decision process for producing a work plan for a production process, and an assistance system for carrying out this method.
Work planning as a sub-area of production scheduling comprises all individually occurring planning measures which, under continual scrutiny of economic efficiency, ensure that a product or a service is fashioned as production requires. Computer-aided work planning is either performed in a specific work planning system or integrated into a CAD/CAM system. In current work planning and CAD/CAM systems, the user is supported by auxiliary functions based on (fixed) formulas and/or rules in the case of decisions which he must make in the course of the work planning. These formulas and/or rules are included a priori in the system and are therefore static during the application, for which reason the individual experience of the user can be taken over into the system only with a very large outlay. Furthermore, the “best” alternative in a specific decision situation is determined with the aid of complex formulas and/or rules, and so the inner logic of a specific decision is frequently difficult for the user to comprehend. The confidence of the user in a decision proposal calculated in this way is therefore low.
In order to be able to let empirical knowledge flow into the work planning process, it is possible to provide work planning systems with functions which expand the systems from the point of view of planning and/or production engineering. Such a system is described, for example, in “Unterstützung der NC-Verfahrenskette durch ein bearbeitungsorientiertes, lernfähiges Technologieplanungssystem” [“Supporting the NC integrated system by a machining-orientated, adaptive technology planning system”] (dissertation by J. E. Burghardt, Karlsruhe 1996). There, the decision process run through during work planning is modelled together with its technological dependencies as a net-like structure, which is capable of machine learning: new empirical knowledge leads to automatic construction of, and/or automatic changes to the network structure. The “best” solution is calculated automatically for each decision situation on the basis of this network structure. However, because of the high degree of complexity and the rules hidden in the network, this way of finding solutions is opaque to the user to a very high degree and credible only under certain conditions. Furthermore, contradictions in the knowledge base can lead to contradictory proposals for solutions, as a result of which the acceptance of such a system can quickly be undermined.
By contrast with the comprehensive rule-based and/or network-based approaches so far described for automated work planning, individual user knowledge in work planning is mostly episodic. Systems which use the method of case-based deduction are particularly suitable for visualizing and making further use of this episodic knowledge in work planning: in this case, already known solutions are being adapted and used to solve a current problem. Furthermore, all new solutions are stored, as a result of which an independent learning system can be implemented. An application of this method of case-based deduction to the work planning process is known, for example, from “Planning for machining workpieces with a partial-order, nonlinear planner” (J. Paulokat and S. We&bgr; in Proceedings of the AAI Fall Symposium, AAI Press 1994). This document presents an example of a very complex work plan which is illustrated as a sequence of steps to be run through. However, the work planning system described here supports only a single decision situation in each case; there is a lack of the modular design and the granularity which would permit the possibility of easy and rapid transfer of the experience obtained in a work plan for the purpose of drawing up a further (other) work plan. The system also lacks functions which would permit detailed planning of each individual work step (for example, rotational speeds and feed rates in a drilling operation).
It is therefore the object of the present invention to develop a method for supporting work planning on the basis of case-based deduction such that the user is presented with proposals for individual decisions at different planning stages, and that it is possible for proposed solutions and empirical knowledge to be transferred directly between different work plans to be set up. It is also the object of the invention to provide a work planning assistance system for carrying out this method.
The object is achieved according to the invention by a method and system for supporting a multistage decision process for computer-aided production of a work plan for a production process in the course of which the aim is to implement technical features on a workpiece. The decision process comprises individual decisions building on one another hierarchically. Each desired feature to be implemented on the workpiece is automatically classified with reference to defined properties. Reference cases, which have properties similar to the desired feature, are selected from a relational case database in accordance with a previously defined similarity measure. The reference cases are automatically evaluated in accordance with the similarity measure and presented to the user for selection together with production operations connected in the respective reference case to the production of the feature. The user makes a decision by selecting one of the proposed reference cases or interactively storing a new precedent. In relation to each production operation, hierarchical decision trees consisting of individual decisions are automatically run through in accordance with the decision made by the user—starting from the production operations associated with this selection, suitable cases being selected from the case database in each decision stage of the decision tree being evaluated in accordance with a similarity measure corresponding to the relevant individual decision, and presented to the user for selection. Each individual decision taken by the user is stored in the case database.
Accordingly, all individual decisions which are made during various work planning processes are stored as cases in a relational database, the so-called “case database”, and combined with one another there. Each decision process to be run through in order to produce a specific work plan is in this case a sequence (or a plurality of sequences) of individual decisions building on one another hierarchically, it being possible for the individual decisions to relate to very different questions (for example, “Which individual operations are necessary to produce a threaded bore of 3 mm diameter?” or “Which feed rate is to be selected when drilling a hole of diameter 2 mm into a steel workpiece?” or “Which clamping accessories are required in the present machining situation?”). Because of the fact that the individual decisions can build on one another in a hierarchical decision chain, and thus that an individual decision simultaneously constitutes a precondition for other decisions, the separately stored individual decisions are combined with one another. This combination of the individual decisions is mapped in the case database as a relational combination of the individual cases. The combination of the cases in the relational case database permits a quick and effective search for the consequences of a decision on further decisions by searching the case database in the sequence of the combinations. The cases stored in the database for a specific decision are presented to the user as solution proposals in this decision situation. The user then has the option of selecting one of these solution proposals, or of producing a new solution not so far contained in the system. In order to
Gerken Harald
Weyrich Michael
Crowell & Moring LLP
Daimler-Chrysler AG
Masinick Michael D.
Picard Leo
LandOfFree
Method and assistance system for supporting work planning... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and assistance system for supporting work planning..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and assistance system for supporting work planning... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3116870