Expert system and method for batch production scheduling and planning

Claims

We claim:

1. A method for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, the parts having a plurality of delivery dates and the parts and parts processor having a plurality of production constraints variable during manufacture, comprising:

creating a knowledge base of select characteristics of the parts processor and parameters of the parts;

generating a plurality of rules expressing a scheduling and planning strategy that substantially satisfies parts delivery dates and meets the production constraints;

preparing parts for manufacture by the parts processor;

combining parts suitable for simultaneous processing by the parts processor into all possible preferred combinations by applying a first plurality of said rules to said knowledge base;

scheduling said preferred combinations for manufacture in batches by applying a second plurality of said rules to said knowledge base;

collecting a plurality of said prepared parts in a first batch and loading said first batch in the parts processor;

operating said parts processor to at least partially manufacture said first batch and thereafter inspecting said first batch for acceptable manufacture; and,

manufacturing the remaining said preferred combinations in said scheduled batches, including the steps of repeating said step of combining parts suitable for simultaneous processing and said step of scheduling said preferred combinations where said manufacture of parts is unacceptable.

2. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 1, wherein said step of generating a plurality of rules includes the step of generating a plurality of rules expressing a multiple-pass scheduling and planning strategy that additionally substantially maximizes use of the parts processor, substantially maximizes part throughput and substantially minimizes energy utilization of the parts processor.

3. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 2, wherein said step of combining parts includes the step of grouping the parts into components based upon said parameters preselected for identifying parts suitable for simultaneous manufacture by the parts processor.

4. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 3, wherein said step of combining parts further includes the step of collecting components into said preferred combinations based upon said parameters preselected for identifying components suitable for simultaneous manufacture by the parts processor and said characteristics preselected for maximizing utilization of the parts processor.

5. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 4, wherein said step of combining parts further includes the step of verifying said parameters in said knowledge base before said step of grouping the parts and said step of collecting components, said parameters including the readiness of each part for further processing.

6. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 5, wherein said step of combining parts further includes the step of assigning a priority to each part indicative of the urgency with which further processing of that part is required.

7. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 6, wherein parts are manufactured in at least one of a plurality of parts processors each having its own characteristics, and further including the step of selecting parts of the highest remaining priority prior to said step of grouping the parts and said step of collecting components.

8. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 7, wherein a first production constraint is that each parts processor is available at different and varying times, a second production constraint is that each part may be manufactured in only select part processors identified in said knowledge base, and said step of collecting components includes the step of collecting for one parts processor at a time, in a preselected order, components into said preferred combinations.

9. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 2, wherein said step of scheduling said preferred combinations for manufacture in batches includes the step of generating heuristically an optimal short-term schedule for the manufacture of parts.

10. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 9, wherein one of the production constraints is that the parts processor is available for manufacture only at certain varying times, and said step of generating heuristically an optimal short-term schedule includes the step of updating the time the parts processor is available for manufacture during the short-term.

11. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 10, wherein another of the production constraints is that at least one tool is required for manufacturing certain of the parts, and said step of generating heuristically an optimal short-term schedule further includes the step of updating the availability of all said tools.

12. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 11, wherein said step of generating heuristically an optimal short-term schedule further includes the step of selecting the next said preferred combination for which the required said tools are available.

13. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 12, wherein the step of generating heuristically an optimal short-term schedule further includes the steps of calculating the time required for manufacturing said selected preferred combination, and scheduling said selected preferred combination if the parts processor is available for sufficient time to manufacture said selected preferred combination.

14. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 13, wherein said step of generating heuristically an optimal short-term schedule further includes the step of repeating said step of combining parts if an additional part has been readied for manufacture during scheduling of said selected preferred combination.

15. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 14, wherein parts are manufactured in a plurality of parts processors each having its own characteristics, a first production constraint is that each parts processor is available at different and varying times, and said step of generating heuristically an optimal short-term schedule further includes the step of determining the next time the availability of any parts processor changes.

16. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 15, wherein a second production constraint is that each part may be manufactured in only select part processors identified in said knowledge base, and said step of generating heuristically an optimal short-term schedule further includes the step of selecting for one parts processor at a time the available preferred combinations until the availability of any parts processor changes or a preselected time limit is reached and the step of repeating said step of combining parts if preferred combinations remain unscheduled after the availability of any parts processor changes or said preselected time limit is reached.

17. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 2, wherein said step of scheduling said preferred combinations for manufacture in batches further includes the step of generating heuristically an optimal long-term schedule for the manufacture of parts.

18. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 17, wherein said step of generating heuristically an optimal long-term schedule further includes the step of updating the time the parts processor is available for manufacture during the long-term.

19. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 18, wherein the parts require partial manufacture other than by the parts processor with preselected time constraints, and said step of generating heuristically an optimal long-term schedule further includes the step of determining the readiness of each part requiring said partial manufacture.

20. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 19, wherein said step of generating heuristically an optimal long-term schedule further includes the step of generating heuristically an optimal short-term schedule for all ready parts, and the step of generating a schedule for said partial manufacture of all parts requiring said partial manufacture.

21. A method for batch scheduling and manufacturing of a plurality of parts, as set forth in claim 20, wherein parts are manufactured in a plurality of parts processors each having its own characteristics, a first production constraint is that each parts processor is available at different and varying times, a second production constraint is that each part may be manufactured in only select part processors identified in said knowledge base, and said step of generating heuristically an optimal long-term schedule further includes the step of inhibiting the scheduling of each part requiring said partial manufacture in all parts processors that do not have sufficient time remaining before they become unavailable to complete said partial processing and manufacture by that part processor.

22. A system for batch scheduling and manufacturing of a plurality of parts having a plurality of delivery dates and a plurality of production constraints variable during manufacture, comprising:

parts processor means for receiving and at least partially manufacturing the parts, said parts processor means having a plurality of production constraints variable during manufacture;

means for preparing the parts for manufacture;

knowledge base means for storing select characteristics of the parts processor and parameters of the parts;

means for interpreting said characteristics and said parameters and generating a multiple-pass schedule to manufacture the parts in batches that substantially satisfies parts delivery dates, substantially maximizes use of the parts processor, substantially maximizes part throughput, substantially minimizes energy utilization of the parts processor and meets the production constraints, said means for interpreting said characteristics and said parameters including

a first plurality of rules for combining parts suitable for simultaneous processing by the parts processor into all possible preferred combinations and

a second plurality of rules for scheduling said preferred combinations for manufacture in batches;

means for receiving said multiple-pass schedule and collecting a plurality of said prepared parts in a first batch, said parts processor means receiving and at least partially manufacturing said first batch; and,

means for inspecting said manufactured first batch for acceptable manufacture;

said parts processor means receiving and at least partially manufacturing the remaining said preferred combinations in said scheduled batches, said means for interpreting generating a further multiple-pass schedule where said manufacture of parts is unacceptable.

23. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 22, wherein said means for interpreting said characteristics and said parameters and generating a schedule includes means for generating a multiple-pass schedule to manufacture parts in batches that additionally substantially maximizes use of the parts processor, substantially maximizes part throughput and substantially minimizes energy utilization of the parts processor.

24. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 23, wherein said first plurality of rules includes a plurality of rules for grouping the parts into components based upon said parameters preselected for identifying parts suitable for simultaneous manufacture by the parts processor.

25. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 24, wherein said first plurality of rules further includes a plurality of rules for collecting components into said preferred combinations based upon said parameters preselected for identifying components suitable for simultaneous manufacture by the parts processor and said characteristics preselected for maximizing utilization of the parts processor.

26. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 25, wherein said first plurality of rules further includes a plurality of rules for verifying said parameters in said knowledge base before grouping the parts and collecting components, said parameters including the readiness of each part for further processing.

27. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 26, wherein said first plurality of rules further includes a plurality of rules for assigning a priority to each part indicative of the urgency with which further processing of that part is required.

28. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 27, further including a plurality of parts processors for manufacturing parts, each said part processor having its own characteristics, and said first plurality of rules further includes a plurality of rules for selecting parts of the highest remaining priority prior to grouping the parts and collecting components.

29. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 28, wherein a first production constraint is that each said parts processor is available at different and varying times, a second production constraint is that each part may be manufactured in only select said part processors identified in said knowledge base, and said rules for collecting components includes a plurality of rules for collecting for one said parts processor at a time, in a preselected order, components into preferred combinations.

30. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 29, wherein each said parts processor is a furnace and said parts are fired in said furnaces.

31. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 23, wherein said second plurality of rules further includes a plurality of rules for generating heuristically an optimal shortterm schedule for the manufacture of parts.

32. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 31, wherein a first production constraint is that the parts processor is available for manufacture only at certain varying times, and said rules for generating heuristically an optimal short-term schedule further includes a plurality of rules for updating the time the parts processor is available for manufacture during the short-term.

33. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 32, wherein a second production constraint is that at least one tool is required for manufacturing certain of the parts, and said rules for generating heuristically an optimal short-term schedule further includes a plurality of rules for updating the availability of all said tools.

34. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 33, wherein said rules for generating heuristically an optimal short-term schedule further includes a plurality of rules for selecting the next said preferred combination for which the required said tools are available.

35. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 34, wherein said rules for generating heuristically an optimal short-term schedule further includes a plurality of rules for calculating the time required for manufacturing said selected preferred combination and scheduling said selected preferred combination if the parts processor is available for sufficient time to manufacture said selected preferred combination.

36. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 35, wherein said rules for generating heuristically an optimal short-term schedule further includes a plurality of rules for repeating said rules for combining parts if an additional part has been readied for manufacture during scheduling of said selected preferred combination.

37. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 36, further including a plurality of parts processors for manufacturing parts, each part processor having its own characteristics, a third production constraint is that each parts processor is available at different and varying times, and said rules for generating heuristically an optimal short-term schedule further includes a plurality of rules for determining the next time the availability of any parts processor changes.

38. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 37, wherein a forth production constraint is that each part may be manufactured in only select part processors identified in said knowledge base, and said rules for generating heuristically an optimal short-term schedule further includes a plurality of rules for selecting for one parts processor at a time the available preferred combinations until the availability of any parts processor changes or a preselected time limit is reached and a plurality of rules for repeating said rules for combining parts if preferred combinations remain unscheduled after the availability of any parts processor changes or said preselected time limit is reached.

39. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 38, wherein each said parts processor is a furnace and said parts are fired in said furnaces.

40. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 23, wherein said rules for scheduling said preferred combinations for manufacture in batches further includes a plurality of rules for generating heuristically an optimal long-term schedule for the manufacture of parts.

41. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 40, wherein said rules for generating heuristically an optimal long-term schedule further includes a plurality of rules for updating the time the parts processor is available for manufacture during the long-term.

42. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 41, wherein the parts require partial manufacture other than by the parts processor with preselected time constraints, and said rules for generating heuristically an optimal long-term schedule further includes a plurality of rules for determining the readiness of each part requiring said partial manufacture.

43. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 42, wherein said rules for generating heuristically an optimal long-term schedule further includes a plurality of rules for generating heuristically an optimal short-term schedule for all ready parts, and a plurality of rules for generating a schedule for said partial manufacture of all parts requiring said partial manufacture.

44. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 43, further including a plurality of parts processors, each said part processor having its own characteristics, a first production constraint is that each said parts processor is available at different and varying times, a second production constraint is that each part may be manufactured in only select said part processors identified in said knowledge base, and said rules for generating heuristically an optimal long-term schedule further includes a plurality of rules for inhibiting the scheduling of each part requiring said partial manufacture in all said parts processors that do not have sufficient time remaining before they become unavailable to complete said partial processing and manufacture by that said part processor.

45. A system for batch scheduling and manufacturing of a plurality of parts by at least one parts processor, as set forth in claim 44, wherein each said parts processor is a furnace and said parts are fired in said furnaces.

TECHNICAL FIELD

The present invention relates generally to computer expert systems. More particularly, the present invention relates to computer expert systems for heuristically optimal scheduling and planning of manufacturing where multiple parts are manufactured simultaneously in multiple-pass batch operations and a large number of manufacturing constraints vary during such manufacture.

BACKGROUND ART

Scheduling and planning manufacturing operations are most desirably done in such manner as to meet required completion dates while simultaneously optimizing throughput and minimizing manufacturing costs. Computers operating under the control of conventional programming have been utilized advantageously to develop optimal manufacturing (also known as production) schedules for many production processes. However, in some systems, particularly those in which a number of discrete parts are manufactured simultaneously in a single operation, the production requirements create combinatorial optimization problems which require an impractical amount of computer time to solve. Additionally, certain production operations require repetitive and variable processing of manufactured parts (in other words, are dynamic in nature). In these instances, conventional programming is incapable of developing production schedules because most of the scheduling actions are non-quantifiable heuristics.

Expert systems, a class of so-called artificial intelligence imbued computer systems in which the analytical skills of an expert are set forth as rules dependent upon and applied against the expert's domain of knowledge (the "knowledge base") are capable of finding solutions to such problems. Several different rule-based expert system paradigms have been utilized in production scheduling and planning. Logic programming has been shown to be applicable in production planning in the paper by W. I. Bullers, S. Y. Nof and A. B. Whinston entitled "Artificial Intelligence in Manufacturing Planning and Control," AIIE Transactions, December 1980, pp 351-363. Similarly, a job shop scheduling system may be based on a schema representation paradigm as disclosed in the paper by M. S. Fox, et. al., entitled "ISIS: A Constraint Directed Reasoning Approach to Job Shop Scheduling," Intelligent Systems Laboratory, The Robotics Institute, Carnegie Mellon University Technical Report, Jun. 21, 1983, pp 83-88. Advantages of using a rule-based system for scheduling production in a so-called flexible manufacturing system have been described in the article by G. Bruno, A. Elia, and P. Laface entitled " A Rule-Based System to Schedule Production," Computer, July 1986, pp 32-40.

Heretofore, to our knowledge, no one has developed the heuristics and knowledge base necessary to produce a complete rule-based expert system capable of scheduling a multiple-pass batch production process, such as the glassing and furnacing operation for manufacture of glass-lined vessels used in chemical manufacturing. Such an expert system must account for numerous constraints including, but not limited to, availability and capacity of the furnace, availability of firing tools, allowed part mix, firing temperature and allowed differences in the thickness of the parts. Such an expert system must also provide the best solution to meeting delivery date requirements, optimizing furnace utilization and minimizing energy consumption. Furthermore, the solution should preferably be optimized for a plurality of furnaces that may operate simultaneously.

DISCLOSURE OF INVENTION

It is, therefore, an object of the invention to provide a rulebased expert system for the batch production planning and scheduling of parts requiring repetitive and variable processing.

It is another object of the invention to provide a rule-based expert system, as above, in which the parts are processed in a parts processor, and the determined schedule satisfies parts delivery dates, maximizes use of the parts processor, maximizes total part throughput, minimizes energy utilization and meets related production constraints.

It is still another object of the invention to provide a rule-based expert system, as above, in which parts and parts processors may be easily added, deleted and modified.

It is yet another object of the invention to provide a rule-based expert system, as above, in which scheduling is optimized for a plurality of parts processors.

It is a further object of the invention to provide a rule-based expert system, as above, in which the parts processor is a furnace and the determined schedule optimizes the firing of parts in a plurality of available furnaces.

These and other objects and advantages of the present invention over existing prior art forms will become more apparent and fully understood from the following description in conjunction with the accompanying drawings.

In general, a method for batch scheduling the manufacture of a plurality of parts by at least one parts processor, the parts having a plurality of delivery dates and the parts and parts processor having a plurality of production constraints variable during manufacture, includes creating a knowledge base of select characteristics of the parts processor and parameters of the parts, and generating a plurality of rules expressing a planning and scheduling strategy that substantially satisfies parts delivery dates and meets the production constraints. Parts suitable for simultaneous processing by the parts processor are combined into all possible preferred combinations by applying a first plurality of rules to the knowledge base. Preferred combinations are scheduled for manufacture in batches by applying a second plurality of rules to the knowledge base.

An expert system for batch scheduling the manufacture of a plurality of parts by at least one parts processor, the parts having a plurality of delivery dates and the parts and parts processor having a plurality of production constraints variable during manufacture, includes a knowledge base for storing select characteristics of the parts processor and parameters of the parts, and means for interpreting the characteristics and parameters and generating a schedule to manufacture the parts in batches that substantially satisfies parts delivery dates and meets the production constraints. The means for interpreting the characteristics and the parameters include a first plurality of rules for combining parts suitable for simultaneous processing by the parts processor into all possible preferred combinations and a second group of rules for scheduling the preferred combinations for manufacture in batches.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram of an exemplary overall production process according to the concept of the present invention for the manufacture of glass-lined vessels.

FIG. 2 is a block diagram of an exemplary overall expert system strategy according to the concept of the present invention for optimally scheduling the production process depicted in FIG. 1.

FIG. 3 is a flow chart showing the grouping of parts to be scheduled for manufacturing into components and the collection of components into preferred combinations.

FIG. 4 is a flow chart showing the short-term scheduling of preferred combinations for firing in a single furnace.

FIG. 5 is a flow chart showing the long-term scheduling of preferred combinations for firing in a single furnace and sandblasting of parts.

FIG. 6 is a flow chart showing the short-term scheduling of preferred combinations for firing in a plurality of furnaces.

FIG. 7 is a block diagram of an exemplary overall expert system according to the concept of the present invention.

PREFERRED EMBODIMENT FOR CARRYING OUT THE INVENTION

FIG. 1 depicts an exemplary overall production process, indicated by the numeral 10, for an exemplary part requiring repetitive production processing, such as glass-lined vessels used for chemical manufacture. The overall production process is begun with an entry of an order 11 for one or more part(s) 12 and fabrication 13 of the ordered part 12. In the present example ordered parts 12 include vessel bodies (sometimes referred to as tanks) of five different size ranges (very large, large, medium, small and very small), agitators (in two sizes, large and small), and miscellaneous small parts including covers, man-holes and supporting legs. Fabrication of different parts 12 require different machining operations and fabrication times. After fabrication, all parts are accumulated in a temporary storage area 14 for further processing.

In order to properly adhere the glass-linings to part 12, the parts 12 must first undergo sandblasting 15 to remove existing oxidation and thereafter must be promptly glassed and fired before substantial oxidation can reoccur. The sandblasted parts are passed through a glassed parts accumulator 20 (hereinafter called "GPA") to a glassing operation 21, and again pass through GPA 20 to a furnace 22 for furnace firing, cool down and inspection 23 (hereinafter called "firing"). Each part undergoes several glassing coats and therefore returns to GPA 20 and glassing 21 as required. Once the final glass coat of a part 12 has been fired and passed inspection, the part 12 is held in a finished parts accumulator 24 (hereinafter called "FPA") until all parts 12 in its associated order are collected and passed to shipping 25. If a part fails to pass any inspection, it may be returned for sandblasting and the entire glassing process begun anew.

FIG. 2 presents an overall expert system strategy for optimally scheduling the production process depicted in FIG. 1. The principal goal of this expert system is to schedule sandblasting 15 and firing 23 of parts 12 to satisfy delivery dates, maximize furnace utilization, maximize parts throughput, minimize energy use and meet production constraints. However, to best understand this strategy, several additional aspects of production process 10 must first be appreciated.

The nature of the production process for manufacture of glasslined vessels, like that of other production processes where parts must be repetitively and variably processed, inherently places numerous constraints on scheduling. Sandblasting, loading and firing operations impose scheduling limitations that may be classified as capacity, resource, time and operations related constraints.

Capacity constraints on scheduling are governed by the characteristics of the furnace 22 and parts 12. Often the characteristics of the furnace 22 and parts 12 permit several parts 12 to be simultaneously fired in the furnace 22. Parts requiring identical processing may be grouped into what may be called "components" for simultaneous firing, although it should be understood that single larger parts or single parts with unique characteristics may themselves make up a component. Components may be collected into combinations preferred for meeting the principal goal (called "preferred combinations").

In the particular example herein each part has been classified into the following categories based upon its type, size and weight:

TABLE 1 ______________________________________ PART CLASSIFICATION Category Part Description ______________________________________ A Very Large Vessel B Large Vessel C Medium Vessel D Small Vessel E Very Small Vessel F Large Agitator G Small Agitator H Cover/Man-Hole/Leg ______________________________________

Since more than one part 12 classified in categories D through H will fit into furnace 22 at one time, multiple parts in each category suitable for identical processing are grouped into components. Specifically, a D component may include no less than 2 and no more than 3 small vessels (category D parts), an E component may include no less than 4 and no more than 8 very small vessels (category E parts), and F, G and H components include no less than 4, 8 and 4 and no more than 10, 15 and 8 parts, respectively, of their own category. In the present example parts suitable for identical processing will have the same temperature and firing cycle (explained hereinbelow) and differences in part thickness within the selected 4 millimeter limitation.

Components may be further combined preferentially based on furnace utilization as follows:

TABLE 2 ______________________________________ GROUP PREFERRED COMBINATIONS Component Component Group # Rank Content Group # Rank Content ______________________________________ 1 1 A 16 1 E & E 2 1 B & B 17 5 E 3 4 B 18 3 E & F 4 2 B & C 19 4 E & G 5 3 B & D 20 4 E & H 6 1 C & C & C 21 1 F & F 7 3 C & C 22 5 F 8 5 C 23 2 F & G 9 4 C & D 24 4 F & H 10 1 D & D 25 1 G & G 11 5 D 26 5 G 12 4 D & E 27 4 G & H 13 3 D & F 28 1 H & H 14 4 D & G 29 5 H 15 4 D & H ______________________________________

Resource constraints on scheduling result from the furnace not always being available (either because it is in use or shut down for maintenance or other reason) and the tools used for part handling (such as racks and other firing tools) not always being available (because either the number of tools is limited or the tools are in use). Time constraints on scheduling include the time remaining for firing before the next shut down period is to begin, the time required for sandblasting each part 12, the time required for firing each part 12, and the time necessary between each firing of each coat for part 12 handling and coat application.

Scheduling constraints resulting from operations (called "operational constraints") may include: any requirement that certain parts 12 always be fired within fixed time intervals; permissible combinations of parts 12 (noted hereinabove); and required firing temperatures and firing cycles (the temperatures at which parts 12 may be loaded into, fired and unloaded from furnace 22). In the present example furnace 22 has three possible firing cycles: a "normal-fire cycle" in which parts 12 are both loaded into and unloaded from furnace 22 at relatively high temperature; a "slow-heat-slow-cool cycle" in which parts 12 are both loaded into and unloaded from furnace 22 at relatively low temperature; and a "mini-cycle" in which parts 12 are loaded into furnace 22 at relatively high temperature and unloaded from furnace 22 at relatively low temperature.

As seen in FIG. 2 the overall expert system scheduling strategy employed is straightforward. Initially a database of parts 12 is created and updated in step 30 to reflect the present status of all parts 12. The database may include the customer order number, a part description, the delivery date, the parts 12 weight, thickness and category, the next glassing sequence to be performed and whether the part 12 is ready for its next glassing sequence. Data maybe entered by an expert system operator 31 acting through a menu-driven interface 32, including video display and keyboard.

In the next major step 33, available parts 12 are grouped into components and the components are collected into all possible preferred combinations. The optimal scheduling for firing 22 and sandblasting 15 is determined in step 34 using the rules set forth hereinafter. If during the implementation of a schedule parts 12 become ready for firing, step 33 is repeated including these parts. Finally, the highly variable nature of the production process 10 requires that the schedule be confirmed either manually or by a real time part status monitor. In the present instance, the schedule is confirmed in step 35 by the system operator 31.

FIG. 3 depicts in more detail the step 33 in which parts are grouped into components and the components are collected into preferred combinations. In step 40 the part database is initialized. Initialization involves: loading the parts data file, initialization file and any file created by prior scheduling; and checking the accuracy of and completing part data in the part data file. The initialization file includes parameters describing allowed categories, components and preferred combinations, preferred combination ranks, times required to sandblast parts in each category, glassing coat sequences, glassing coats and their corresponding firing temperatures, and firing tools available for each category.

Once the part database is initialized, the readiness of each part for its next firing is found in step 41 by the comparison of the current time with the next "ready time" associated in the database field for each part 12. A database field for each part is then assigned an appropriate attribute such as "Y" to indicate part 12 readiness or "N" to indicate not ready.

Next, each part 12 is assigned a priority in step 42 to indicate the urgency with which firing of its next coat is required. Priority rating is determined by the need to meet the delivery deadlines and avoid part oxidation after sandblasting and the first two ground coats. The rules for priority assignment for the first three coats are based on preselected maximum allowable times between sandblasting and the first ground coat, the first ground coat and the second ground coat, and the second ground coat and the first cover coat. For example, one such rule may take the following form:

IF: The next coat for this part is the first ground coat,

AND: This is the first day after sandblasting,

AND: The next day is fully available,

THEN: Assign priority 2 ,

ELSE: Assign priority 1 (the highest priority).

Rules also check the shutdown schedule to determine which days are fully available for firing since it is critical that the first three coats are completed within specified time limits.

After the part has undergone the first three coats, priority may be assigned based upon the time between the promised and approximate delivery dates (i.e., the approximate date part 12 is delivered to FPA 24). The difference between the promised and approximate delivery dates is computed for each part. The approximate delivery date may be obtained assuming a fixed amount of time for firing of each coat. Of course, it will be recognized that priority assignment may be readily varied as desired to a point where the permissible time between sandblasting and each coat and the time required for firing each coat is individually set. Additionally, the number of priority levels may be varied to suit the application.

Once readiness of each part 12 has been checked and priority assigned, ready parts 12 are grouped in step 43 into components in accordance with the rules embodying the above explained considerations. As ready parts 12 are grouped into components, parts with the highest priority are selected first from all eligible candidates, and parts with lower priority are selected next until all ready parts 12 have been assigned. The maximum and minimum part 12 thickness is determined for each component and retained for use in collecting preferred combinations.

Preferred combinations of components maximizing furnace utilization are set forth above in Table 2 together with their rank. Table 2 indicates that greater furnace utilization occurs when two category B components are fired together than when one category B component is fired, and so on. During component collection, rules insure that only components having compatible firing cycles, firing temperatures, and thickness ranges are collected, and components are combined in order of descending priority. The rank for each preferred combination is retained for use in scheduling.

Once all possible preferred combinations have been found, scheduling of the preferred combinations for firing takes place in step 34. In fact, scheduling is done in two parts, selectable by operator 31: long-term and short-term.

Long-term scheduling is used to generate weekly, biweekly, or monthly schedules to enable operator 31 to plan equipment utilization, to understand possible delivery dates, and to schedule future shut downs at the most appropriate times. Long-term scheduling also generates a sandblasting schedule based on the demand from the furnacing operation. In this way, the sandblasting 15 operation is driven by the furnacing 23 operation, and therefore all parts 12 after sandblasting 15 will have a greater likelihood of going through the furnacing 23 operation before oxidation occurs. However, since long-term scheduling cannot account for future new part orders and future distu